Annex_2_CNN_approach_results_analyzis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--NOTEBOOK_INFORMATION-->\n",
    "<img id=\"r-1060983\" data-claire-element-id=\"1061343\" src=\"http://www.siteduzero.com/favicon.ico\" alt=\"Image utilisateur\">\n",
    "    <p>\n",
    "        **<font color='#D2691E'size=\"6\">Image classification ANNEX (2/2)</font>**.\n",
    "    </p>\n",
    "    <p>\n",
    "        This notebook analyzes the results from different neural networks learning (notebook \"Annex_1_CNN_approach_transfer_learning_over_parameters\").\n",
    "    </p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p>\n",
    "    <center>\n",
    "        **<font color='\t#D2691E'size=\"6\">PLAN</font>**\n",
    "    </center>\n",
    "</p>\n",
    "\n",
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">0) Libraries, functions and datasources import</font>**\n",
    "</p>\n",
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">I) Preprocessing over the results</font>**\n",
    "</p>\n",
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">II) Vizualizing the accuracy depending on the parameters</font>**\n",
    "</p>\n",
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">III) Vizualizing the losses and accuracies depending on the training epochs</font>**\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">0) Libraries, functions and datasources import</font>**\n",
    "</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from functions import convert_in_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>races</th>\n",
       "      <th>training_len</th>\n",
       "      <th>testing_len</th>\n",
       "      <th>iter_number</th>\n",
       "      <th>n_races</th>\n",
       "      <th>random_selection_number</th>\n",
       "      <th>batch_size</th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>fitting_time</th>\n",
       "      <th>prediction_time</th>\n",
       "      <th>epochs_losses</th>\n",
       "      <th>epochs_accuracies</th>\n",
       "      <th>epochs_val_losses</th>\n",
       "      <th>epochs_val_accuracies</th>\n",
       "      <th>test_loss</th>\n",
       "      <th>test_accuracy</th>\n",
       "      <th>is_augmented</th>\n",
       "      <th>optimizer</th>\n",
       "      <th>learning_rate_categ</th>\n",
       "      <th>run</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>3077</td>\n",
       "      <td>125</td>\n",
       "      <td>[11.177390171632402, 5.857689602159346, 2.6551...</td>\n",
       "      <td>[0.18174409032013714, 0.5069274632252613, 0.74...</td>\n",
       "      <td>[15.50204170040968, 15.00998678440001, 12.2796...</td>\n",
       "      <td>[0.0, 0.029268292483033205, 0.2048780423838918...</td>\n",
       "      <td>3.816269</td>\n",
       "      <td>0.722826</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000089</td>\n",
       "      <td>3454</td>\n",
       "      <td>122</td>\n",
       "      <td>[10.441518563610224, 5.4293057353105985, 3.884...</td>\n",
       "      <td>[0.23553382233088835, 0.5973920026443407, 0.71...</td>\n",
       "      <td>[15.783933965171256, 12.326112328506097, 12.78...</td>\n",
       "      <td>[0.002439024335727459, 0.19268292479398774, 0....</td>\n",
       "      <td>3.838663</td>\n",
       "      <td>0.733696</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>2</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000443</td>\n",
       "      <td>2209</td>\n",
       "      <td>122</td>\n",
       "      <td>[9.8417969667144, 5.32107679107467, 4.79954731...</td>\n",
       "      <td>[0.31621842245406606, 0.6479217584966739, 0.68...</td>\n",
       "      <td>[12.235719762197355, 11.661158957132479, 11.52...</td>\n",
       "      <td>[0.21219512674866653, 0.2682926887419166, 0.27...</td>\n",
       "      <td>7.138068</td>\n",
       "      <td>0.543478</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>3</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000753</td>\n",
       "      <td>1477</td>\n",
       "      <td>122</td>\n",
       "      <td>[10.426884437054275, 7.350310457364854, 9.7848...</td>\n",
       "      <td>[0.29910350730604723, 0.5264873691662211, 0.38...</td>\n",
       "      <td>[16.118097258777155, 16.118097258777155, 16.11...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>10.832995</td>\n",
       "      <td>0.327899</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>4</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004071</td>\n",
       "      <td>1475</td>\n",
       "      <td>122</td>\n",
       "      <td>[14.32642020184331, 14.633708820949265, 14.633...</td>\n",
       "      <td>[0.09942950298000179, 0.09209453961838526, 0.0...</td>\n",
       "      <td>[16.118097258777155, 16.118097258777155, 16.11...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>15.008516</td>\n",
       "      <td>0.068841</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>5</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005730</td>\n",
       "      <td>1477</td>\n",
       "      <td>122</td>\n",
       "      <td>[14.377608573621332, 14.35784860733766, 14.357...</td>\n",
       "      <td>[0.10268948660721697, 0.10920945234359913, 0.1...</td>\n",
       "      <td>[16.118097258777155, 16.118097258777155, 16.11...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.804120</td>\n",
       "      <td>0.081522</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>6</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000027</td>\n",
       "      <td>1452</td>\n",
       "      <td>113</td>\n",
       "      <td>[12.859274773210181, 9.429001272356553, 6.2099...</td>\n",
       "      <td>[0.10424028090843042, 0.26855123606461095, 0.4...</td>\n",
       "      <td>[15.520961927989173, 15.956519313590237, 16.03...</td>\n",
       "      <td>[0.00793650775911316, 0.0, 0.0, 0.002645502586...</td>\n",
       "      <td>6.095660</td>\n",
       "      <td>0.547059</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>7</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000094</td>\n",
       "      <td>1780</td>\n",
       "      <td>113</td>\n",
       "      <td>[10.854263379801289, 5.383011704198885, 3.2654...</td>\n",
       "      <td>[0.2049469942812785, 0.5927561940658219, 0.738...</td>\n",
       "      <td>[16.11809730529785, 16.115518347926873, 16.118...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.329826</td>\n",
       "      <td>0.633333</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>8</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000462</td>\n",
       "      <td>1361</td>\n",
       "      <td>112</td>\n",
       "      <td>[9.503772049826363, 6.354559300223846, 6.12258...</td>\n",
       "      <td>[0.34098940186018234, 0.5697879856551072, 0.60...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>8.239412</td>\n",
       "      <td>0.484314</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>9</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000946</td>\n",
       "      <td>1362</td>\n",
       "      <td>113</td>\n",
       "      <td>[10.594130981094846, 9.908407009111276, 10.101...</td>\n",
       "      <td>[0.2985865757488004, 0.3780918724755938, 0.370...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>12.452109</td>\n",
       "      <td>0.227451</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>10</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.001557</td>\n",
       "      <td>1364</td>\n",
       "      <td>113</td>\n",
       "      <td>[11.499933091153526, 10.687108548706917, 11.12...</td>\n",
       "      <td>[0.256183743160942, 0.3321554788747019, 0.3074...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>12.548790</td>\n",
       "      <td>0.219608</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>11</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.008522</td>\n",
       "      <td>1367</td>\n",
       "      <td>112</td>\n",
       "      <td>[14.262176338438431, 14.409464812531487, 14.40...</td>\n",
       "      <td>[0.10689045948242973, 0.10600706758537057, 0.1...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.822327</td>\n",
       "      <td>0.080392</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>12</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000021</td>\n",
       "      <td>1471</td>\n",
       "      <td>117</td>\n",
       "      <td>[11.025774697125968, 7.274011296741033, 4.4514...</td>\n",
       "      <td>[0.17796610106351013, 0.413559321629799, 0.605...</td>\n",
       "      <td>[15.697799256610386, 16.098241999669728, 16.09...</td>\n",
       "      <td>[0.007614213027778616, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.951484</td>\n",
       "      <td>0.566038</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>13</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000090</td>\n",
       "      <td>1418</td>\n",
       "      <td>120</td>\n",
       "      <td>[10.217060638686357, 5.400305562100168, 3.1394...</td>\n",
       "      <td>[0.25084745947082165, 0.6008474579301931, 0.77...</td>\n",
       "      <td>[16.111351148731213, 16.11809730529785, 16.118...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>6.159383</td>\n",
       "      <td>0.592453</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>14</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000157</td>\n",
       "      <td>1437</td>\n",
       "      <td>118</td>\n",
       "      <td>[9.837554091114109, 3.447510222257194, 1.59543...</td>\n",
       "      <td>[0.28983050859454323, 0.7237288113367759, 0.86...</td>\n",
       "      <td>[16.117790386761506, 16.11809730529785, 16.118...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.221603</td>\n",
       "      <td>0.652830</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>15</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000587</td>\n",
       "      <td>1427</td>\n",
       "      <td>118</td>\n",
       "      <td>[10.605151224944551, 8.833039421146198, 8.9157...</td>\n",
       "      <td>[0.31101694488424364, 0.44830507646172735, 0.4...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>10.933599</td>\n",
       "      <td>0.316981</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>16</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.003289</td>\n",
       "      <td>1433</td>\n",
       "      <td>117</td>\n",
       "      <td>[13.438748537483862, 13.53646965350135, 13.536...</td>\n",
       "      <td>[0.15677966101694915, 0.16016949195477923, 0.1...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.202171</td>\n",
       "      <td>0.118868</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.006520</td>\n",
       "      <td>1463</td>\n",
       "      <td>121</td>\n",
       "      <td>[14.52452774371131, 14.56092522508007, 14.5609...</td>\n",
       "      <td>[0.09406779649651657, 0.09661016941575681, 0.0...</td>\n",
       "      <td>[16.11809730529785, 16.11809730529785, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.932047</td>\n",
       "      <td>0.073585</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>['bluetick', 'border_collie', 'border_terrier'...</td>\n",
       "      <td>1495</td>\n",
       "      <td>505</td>\n",
       "      <td>18</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000022</td>\n",
       "      <td>1697</td>\n",
       "      <td>111</td>\n",
       "      <td>[12.375398138184083, 9.79439748513922, 5.40574...</td>\n",
       "      <td>[0.12845673432688326, 0.2542372876304691, 0.52...</td>\n",
       "      <td>[15.606216185870656, 15.482723521676293, 15.52...</td>\n",
       "      <td>[0.013368984156432636, 0.029411764546511645, 0...</td>\n",
       "      <td>5.076590</td>\n",
       "      <td>0.619802</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>['bluetick', 'border_collie', 'border_terrier'...</td>\n",
       "      <td>1495</td>\n",
       "      <td>505</td>\n",
       "      <td>19</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000095</td>\n",
       "      <td>1348</td>\n",
       "      <td>111</td>\n",
       "      <td>[12.220591681222635, 7.596678482433391, 5.1738...</td>\n",
       "      <td>[0.17395182801983805, 0.4683318553919456, 0.62...</td>\n",
       "      <td>[16.110363363582184, 16.11809730529785, 16.118...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>6.485281</td>\n",
       "      <td>0.562376</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1672</td>\n",
       "      <td>90</td>\n",
       "      <td>[9.61944356171981, 5.981166259102199, 5.565419...</td>\n",
       "      <td>[0.35000000424359157, 0.6086956586526788, 0.64...</td>\n",
       "      <td>[14.36760568774096, 15.284007802460016, 13.127...</td>\n",
       "      <td>[0.09120521211468824, 0.04560260605734412, 0.1...</td>\n",
       "      <td>7.928827</td>\n",
       "      <td>0.497585</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1096</td>\n",
       "      <td>90</td>\n",
       "      <td>[11.11389104179714, 6.314466725225034, 4.44965...</td>\n",
       "      <td>[0.18043478205800056, 0.5271739098040954, 0.67...</td>\n",
       "      <td>[16.085105100749757, 16.107627346771935, 16.10...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>6.912676</td>\n",
       "      <td>0.533816</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000428</td>\n",
       "      <td>1117</td>\n",
       "      <td>90</td>\n",
       "      <td>[13.487566305243451, 13.388840011928393, 14.05...</td>\n",
       "      <td>[0.14673913359318091, 0.16847826159842635, 0.1...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.560792</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000428</td>\n",
       "      <td>1639</td>\n",
       "      <td>90</td>\n",
       "      <td>[7.219512410785841, 2.540557677331178, 1.89829...</td>\n",
       "      <td>[0.43913043517133465, 0.8195652106533879, 0.85...</td>\n",
       "      <td>[15.243229061462204, 13.201828611790162, 13.02...</td>\n",
       "      <td>[0.026058631339368293, 0.17915309834557946, 0....</td>\n",
       "      <td>5.245852</td>\n",
       "      <td>0.659420</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000770</td>\n",
       "      <td>1122</td>\n",
       "      <td>90</td>\n",
       "      <td>[12.506660606550133, 13.595264476278555, 13.52...</td>\n",
       "      <td>[0.21413043538189452, 0.15652173749454643, 0.1...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.171466</td>\n",
       "      <td>0.120773</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000770</td>\n",
       "      <td>1095</td>\n",
       "      <td>89</td>\n",
       "      <td>[10.682513879693072, 6.306996697964876, 5.0812...</td>\n",
       "      <td>[0.29021738971705024, 0.5749999952057133, 0.67...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>9.279782</td>\n",
       "      <td>0.420290</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>1402</td>\n",
       "      <td>90</td>\n",
       "      <td>[13.795615154763926, 13.630303527997887, 13.56...</td>\n",
       "      <td>[0.14021739128815092, 0.15434782547147377, 0.1...</td>\n",
       "      <td>[13.545501460469895, 13.387995778931858, 13.38...</td>\n",
       "      <td>[0.15960912362760363, 0.1693811012789947, 0.16...</td>\n",
       "      <td>13.159218</td>\n",
       "      <td>0.183575</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>1094</td>\n",
       "      <td>90</td>\n",
       "      <td>[13.70452408168627, 14.068295706873355, 14.068...</td>\n",
       "      <td>[0.13804347845523254, 0.12717391339981038, 0.1...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.560792</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005693</td>\n",
       "      <td>1115</td>\n",
       "      <td>90</td>\n",
       "      <td>[13.924330690632695, 14.033256157584812, 14.03...</td>\n",
       "      <td>[0.12934782598977504, 0.12934782598977504, 0.1...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.560792</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005693</td>\n",
       "      <td>1093</td>\n",
       "      <td>90</td>\n",
       "      <td>[13.995241020036781, 14.068295706873355, 14.06...</td>\n",
       "      <td>[0.12391304159941881, 0.1271739136427641, 0.12...</td>\n",
       "      <td>[16.11809726180782, 16.11809726180782, 16.1180...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.560792</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000055</td>\n",
       "      <td>1485</td>\n",
       "      <td>88</td>\n",
       "      <td>[12.531751771991173, 9.042676127358769, 5.7434...</td>\n",
       "      <td>[0.10561798094363695, 0.3067415812042322, 0.52...</td>\n",
       "      <td>[15.782424457948215, 16.034975787204523, 16.10...</td>\n",
       "      <td>[0.003367003263523081, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.981250</td>\n",
       "      <td>0.570000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000291</td>\n",
       "      <td>2319</td>\n",
       "      <td>88</td>\n",
       "      <td>[12.410123653626174, 9.598922086565683, 7.7693...</td>\n",
       "      <td>[0.19550561603535427, 0.3943820254186566, 0.51...</td>\n",
       "      <td>[14.268121334037396, 14.528891213413843, 13.96...</td>\n",
       "      <td>[0.11111111552626998, 0.09427609467747236, 0.1...</td>\n",
       "      <td>10.516670</td>\n",
       "      <td>0.342500</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000291</td>\n",
       "      <td>1580</td>\n",
       "      <td>88</td>\n",
       "      <td>[11.956391655996944, 5.9083209198512385, 3.571...</td>\n",
       "      <td>[0.15168539309099818, 0.5415730302253466, 0.74...</td>\n",
       "      <td>[13.86413397612395, 14.419594356909345, 14.405...</td>\n",
       "      <td>[0.12794612634061564, 0.08080807900187945, 0.1...</td>\n",
       "      <td>5.823327</td>\n",
       "      <td>0.607500</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000604</td>\n",
       "      <td>1632</td>\n",
       "      <td>88</td>\n",
       "      <td>[13.85870252030619, 13.854321597667223, 13.854...</td>\n",
       "      <td>[0.12134831557782848, 0.14044943636053064, 0.1...</td>\n",
       "      <td>[16.118099874117558, 16.118099874117558, 16.11...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.425695</td>\n",
       "      <td>0.105000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000604</td>\n",
       "      <td>2604</td>\n",
       "      <td>88</td>\n",
       "      <td>[10.893494188115838, 5.097742683432076, 3.2496...</td>\n",
       "      <td>[0.22921348605932815, 0.6415730279483153, 0.76...</td>\n",
       "      <td>[14.632227765992033, 13.788364712236707, 14.00...</td>\n",
       "      <td>[0.05387205266791963, 0.14141413699898253, 0.1...</td>\n",
       "      <td>4.464179</td>\n",
       "      <td>0.710000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>1603</td>\n",
       "      <td>88</td>\n",
       "      <td>[13.505435204238035, 13.836211440268528, 13.83...</td>\n",
       "      <td>[0.14719100890869505, 0.14157303313860733, 0.1...</td>\n",
       "      <td>[16.118099874117558, 16.118099874117558, 16.11...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>14.425695</td>\n",
       "      <td>0.105000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>3042</td>\n",
       "      <td>88</td>\n",
       "      <td>[10.081505084305666, 4.522153259663099, 3.4023...</td>\n",
       "      <td>[0.2853932672002342, 0.6707865274354313, 0.773...</td>\n",
       "      <td>[14.218336965901281, 13.99168165444525, 13.990...</td>\n",
       "      <td>[0.1010101050238818, 0.1245791261846369, 0.131...</td>\n",
       "      <td>5.603336</td>\n",
       "      <td>0.640000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>1640</td>\n",
       "      <td>88</td>\n",
       "      <td>[12.193586542365257, 10.661302737975388, 8.889...</td>\n",
       "      <td>[0.12359550561797752, 0.19101123762934397, 0.2...</td>\n",
       "      <td>[14.465690612792969, 14.751871109008789, 14.88...</td>\n",
       "      <td>[0.033670034259557724, 0.04377104341983795, 0....</td>\n",
       "      <td>8.045590</td>\n",
       "      <td>0.377500</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-6,1e-5]</td>\n",
       "      <td>run_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1457</td>\n",
       "      <td>96</td>\n",
       "      <td>[10.736687477159057, 7.303910214588492, 4.2432...</td>\n",
       "      <td>[0.2033023729966521, 0.4200206394043008, 0.629...</td>\n",
       "      <td>[15.823758878205952, 16.04527726778674, 16.097...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.366906</td>\n",
       "      <td>0.642032</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>104</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1494</td>\n",
       "      <td>95</td>\n",
       "      <td>[12.933459277984651, 12.577445039070058, 11.97...</td>\n",
       "      <td>[0.09803921679348153, 0.12280701971214007, 0.1...</td>\n",
       "      <td>[14.019623103894686, 14.42137057714787, 14.778...</td>\n",
       "      <td>[0.024767801834518325, 0.012383900605857187, 0...</td>\n",
       "      <td>10.915709</td>\n",
       "      <td>0.193995</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>2015</td>\n",
       "      <td>95</td>\n",
       "      <td>[8.19136002730536, 2.224062449422783, 0.796706...</td>\n",
       "      <td>[0.3725490320946779, 0.8266253913027091, 0.933...</td>\n",
       "      <td>[13.19497582639334, 12.653280547528814, 12.879...</td>\n",
       "      <td>[0.14551083554424368, 0.21052631800388774, 0.1...</td>\n",
       "      <td>3.637250</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>3112</td>\n",
       "      <td>97</td>\n",
       "      <td>[12.499718916059402, 9.08367666592908, 5.51035...</td>\n",
       "      <td>[0.10939112539384879, 0.297213626320025, 0.520...</td>\n",
       "      <td>[14.741868503322543, 15.533892785063468, 14.81...</td>\n",
       "      <td>[0.012383900651990814, 0.006191950325995407, 0...</td>\n",
       "      <td>3.831526</td>\n",
       "      <td>0.732102</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>107</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000067</td>\n",
       "      <td>1361</td>\n",
       "      <td>107</td>\n",
       "      <td>[7.601912935451827, 2.8883877705986407, 2.1372...</td>\n",
       "      <td>[0.4334365318311134, 0.7843137323179727, 0.850...</td>\n",
       "      <td>[16.11810039366731, 16.11810039366731, 16.1181...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.459007</td>\n",
       "      <td>0.655889</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>108</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000067</td>\n",
       "      <td>1356</td>\n",
       "      <td>96</td>\n",
       "      <td>[13.238199061776585, 10.050782345396815, 7.420...</td>\n",
       "      <td>[0.1073271394760124, 0.27863777055951955, 0.46...</td>\n",
       "      <td>[15.46262528800374, 16.09364617645925, 16.1085...</td>\n",
       "      <td>[0.006191950325995407, 0.0, 0.0, 0.0]</td>\n",
       "      <td>8.237578</td>\n",
       "      <td>0.475751</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>8158</td>\n",
       "      <td>96</td>\n",
       "      <td>[12.094785657829544, 9.932752012837414, 7.9038...</td>\n",
       "      <td>[0.14447884455369353, 0.2662538664628847, 0.40...</td>\n",
       "      <td>[15.704390956890473, 15.926995829532022, 15.79...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0030959753071652845, 0.04643...</td>\n",
       "      <td>5.026830</td>\n",
       "      <td>0.644342</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>2091</td>\n",
       "      <td>96</td>\n",
       "      <td>[12.661890428501755, 12.380870878142838, 11.91...</td>\n",
       "      <td>[0.10010320072757202, 0.1207430337789258, 0.14...</td>\n",
       "      <td>[15.218670963133821, 15.288978062916097, 15.38...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.006191950614330569]</td>\n",
       "      <td>11.176276</td>\n",
       "      <td>0.168591</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>4922</td>\n",
       "      <td>96</td>\n",
       "      <td>[10.56009166528566, 4.163821252876022, 2.45548...</td>\n",
       "      <td>[0.2569659387364107, 0.6718266078562191, 0.807...</td>\n",
       "      <td>[13.896825524675588, 12.75900555542748, 12.684...</td>\n",
       "      <td>[0.086687306870617, 0.19195046125919826, 0.204...</td>\n",
       "      <td>3.594073</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>5297</td>\n",
       "      <td>96</td>\n",
       "      <td>[12.84661759533011, 11.271038799463042, 9.4459...</td>\n",
       "      <td>[0.10423116778072558, 0.1991744041442871, 0.30...</td>\n",
       "      <td>[13.369037610458516, 14.198510066643587, 13.98...</td>\n",
       "      <td>[0.05572755293395866, 0.03715170195597244, 0.0...</td>\n",
       "      <td>6.137263</td>\n",
       "      <td>0.554273</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>3212</td>\n",
       "      <td>96</td>\n",
       "      <td>[11.282455689397759, 8.86174060242833, 7.52606...</td>\n",
       "      <td>[0.24871000315382755, 0.42208461689506155, 0.5...</td>\n",
       "      <td>[12.769200717701631, 13.259642054791051, 12.75...</td>\n",
       "      <td>[0.1981424196585782, 0.16099071078256189, 0.20...</td>\n",
       "      <td>6.862218</td>\n",
       "      <td>0.568129</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>7417</td>\n",
       "      <td>96</td>\n",
       "      <td>[12.795913973828956, 9.555369752110343, 4.9030...</td>\n",
       "      <td>[0.10216718238573694, 0.2745097998310538, 0.57...</td>\n",
       "      <td>[14.268447320896774, 13.792559396371752, 13.49...</td>\n",
       "      <td>[0.0433436534353085, 0.08359133459097091, 0.12...</td>\n",
       "      <td>3.549291</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000008</td>\n",
       "      <td>1739</td>\n",
       "      <td>98</td>\n",
       "      <td>[11.583652914836023, 8.805376688317896, 7.2445...</td>\n",
       "      <td>[0.16986855689748837, 0.3488372009552646, 0.47...</td>\n",
       "      <td>[15.894092126326127, 16.023284276326496, 16.04...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>7.677936</td>\n",
       "      <td>0.452703</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>13</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000008</td>\n",
       "      <td>2345</td>\n",
       "      <td>99</td>\n",
       "      <td>[12.272719182669212, 11.694719937500988, 10.68...</td>\n",
       "      <td>[0.12841253913750664, 0.14459049290364384, 0.1...</td>\n",
       "      <td>[15.269959218574293, 15.42425502430309, 15.624...</td>\n",
       "      <td>[0.012121211850281918, 0.0030303029625704794, ...</td>\n",
       "      <td>9.369028</td>\n",
       "      <td>0.234234</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>14</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>2393</td>\n",
       "      <td>99</td>\n",
       "      <td>[8.738685776177263, 2.779545922612035, 1.40377...</td>\n",
       "      <td>[0.35389282259830207, 0.7724974700245023, 0.88...</td>\n",
       "      <td>[16.063047351259172, 15.81689707438151, 12.779...</td>\n",
       "      <td>[0.0, 0.006060605925140959, 0.1939393896045106...</td>\n",
       "      <td>3.732816</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>118</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>15</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>1686</td>\n",
       "      <td>98</td>\n",
       "      <td>[11.888713320779367, 8.879410602446152, 6.0340...</td>\n",
       "      <td>[0.13751263845678047, 0.33063700689705366, 0.4...</td>\n",
       "      <td>[15.033767873590643, 16.032680569273055, 16.10...</td>\n",
       "      <td>[0.0121212119913914, 0.0, 0.0, 0.0]</td>\n",
       "      <td>5.818482</td>\n",
       "      <td>0.578829</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>16</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000073</td>\n",
       "      <td>1624</td>\n",
       "      <td>98</td>\n",
       "      <td>[11.166352626889251, 7.717718957528788, 6.3307...</td>\n",
       "      <td>[0.2709807908149291, 0.502527802007384, 0.5965...</td>\n",
       "      <td>[16.11810036861535, 16.11810036861535, 16.1181...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0]</td>\n",
       "      <td>8.026507</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>17</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000073</td>\n",
       "      <td>1815</td>\n",
       "      <td>98</td>\n",
       "      <td>[12.853760399157888, 9.65257440273875, 5.97255...</td>\n",
       "      <td>[0.1213346834400908, 0.2942365967226945, 0.535...</td>\n",
       "      <td>[15.856012170965021, 16.108810771595348, 16.11...</td>\n",
       "      <td>[0.00303030303312522, 0.0, 0.0, 0.0]</td>\n",
       "      <td>6.849744</td>\n",
       "      <td>0.524775</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>18</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>4357</td>\n",
       "      <td>98</td>\n",
       "      <td>[12.051452930824581, 9.182866433030794, 6.9785...</td>\n",
       "      <td>[0.1678463114525358, 0.3053589442742968, 0.431...</td>\n",
       "      <td>[14.675517342307351, 14.236758665605025, 14.17...</td>\n",
       "      <td>[0.03333333202383735, 0.06060606376691298, 0.0...</td>\n",
       "      <td>3.974939</td>\n",
       "      <td>0.707207</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>19</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1672</td>\n",
       "      <td>99</td>\n",
       "      <td>[13.157143132872239, 12.729341613993245, 12.10...</td>\n",
       "      <td>[0.08695652133232422, 0.09302325837532599, 0.1...</td>\n",
       "      <td>[13.869226455688477, 14.43110214580189, 15.068...</td>\n",
       "      <td>[0.048484849658879364, 0.036363635550845756, 0...</td>\n",
       "      <td>11.600196</td>\n",
       "      <td>0.150901</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>4380</td>\n",
       "      <td>99</td>\n",
       "      <td>[10.506150655486104, 5.842451920523803, 4.7496...</td>\n",
       "      <td>[0.26289180596147194, 0.6036400478578072, 0.67...</td>\n",
       "      <td>[16.118100599809125, 16.118100599809125, 16.05...</td>\n",
       "      <td>[0.0, 0.0, 0.0030303031036799603, 0.0, 0.00909...</td>\n",
       "      <td>5.505957</td>\n",
       "      <td>0.650901</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>21</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>1811</td>\n",
       "      <td>135</td>\n",
       "      <td>[13.170345185861308, 10.6054297804953, 7.32566...</td>\n",
       "      <td>[0.08392315420451131, 0.2082912012467611, 0.42...</td>\n",
       "      <td>[14.294045188210227, 15.977744015780361, 16.10...</td>\n",
       "      <td>[0.009090908887711439, 0.0, 0.0, 0.0]</td>\n",
       "      <td>6.114878</td>\n",
       "      <td>0.529279</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>125 rows × 20 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 races  training_len  \\\n",
       "0    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "1    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "2    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "3    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "4    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "5    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "6    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "7    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "8    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "9    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "10   ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "11   ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "12   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "13   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "14   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "15   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "16   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "17   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "18   ['bluetick', 'border_collie', 'border_terrier'...          1495   \n",
       "19   ['bluetick', 'border_collie', 'border_terrier'...          1495   \n",
       "20   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "21   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "22   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "23   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "24   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "25   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "26   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "27   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "28   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "29   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "..                                                 ...           ...   \n",
       "95   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "96   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "97   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "98   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "99   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "100  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "101  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "102  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "103  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "104  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "105  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "106  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "107  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "108  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "109  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "110  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "111  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "112  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "113  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "114  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "115  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "116  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "117  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "118  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "119  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "120  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "121  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "122  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "123  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "124  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "\n",
       "     testing_len  iter_number  n_races  random_selection_number  batch_size  \\\n",
       "0            552            0       12                        0         100   \n",
       "1            552            1       12                        0         100   \n",
       "2            552            2       12                        0         100   \n",
       "3            552            3       12                        0         100   \n",
       "4            552            4       12                        0         100   \n",
       "5            552            5       12                        0         100   \n",
       "6            510            6       12                        1         100   \n",
       "7            510            7       12                        1         100   \n",
       "8            510            8       12                        1         100   \n",
       "9            510            9       12                        1         100   \n",
       "10           510           10       12                        1         100   \n",
       "11           510           11       12                        1         100   \n",
       "12           530           12       12                        2         100   \n",
       "13           530           13       12                        2         100   \n",
       "14           530           14       12                        2         100   \n",
       "15           530           15       12                        2         100   \n",
       "16           530           16       12                        2         100   \n",
       "17           530           17       12                        2         100   \n",
       "18           505           18       12                        3         100   \n",
       "19           505           19       12                        3         100   \n",
       "20           414            0       10                        0         100   \n",
       "21           414            1       10                        0         100   \n",
       "22           414            2       10                        0         100   \n",
       "23           414            3       10                        0         100   \n",
       "24           414            4       10                        0         100   \n",
       "25           414            5       10                        0         100   \n",
       "26           414            6       10                        0         100   \n",
       "27           414            7       10                        0         100   \n",
       "28           414            8       10                        0         100   \n",
       "29           414            9       10                        0         100   \n",
       "..           ...          ...      ...                      ...         ...   \n",
       "95           400            5       10                        0         200   \n",
       "96           400            6       10                        0         200   \n",
       "97           400            7       10                        0         200   \n",
       "98           400            8       10                        0         200   \n",
       "99           400            9       10                        0         200   \n",
       "100          400           10       10                        0         200   \n",
       "101          400           11       10                        0         200   \n",
       "102          400           12       10                        0         300   \n",
       "103          433            0       10                        0         200   \n",
       "104          433            1       10                        0         200   \n",
       "105          433            2       10                        0         200   \n",
       "106          433            3       10                        0         200   \n",
       "107          433            4       10                        0         200   \n",
       "108          433            5       10                        0         200   \n",
       "109          433            6       10                        0         300   \n",
       "110          433            7       10                        0         300   \n",
       "111          433            8       10                        0         300   \n",
       "112          433            9       10                        0         300   \n",
       "113          433           10       10                        0         300   \n",
       "114          433           11       10                        0         300   \n",
       "115          444           12       10                        1         200   \n",
       "116          444           13       10                        1         200   \n",
       "117          444           14       10                        1         200   \n",
       "118          444           15       10                        1         200   \n",
       "119          444           16       10                        1         200   \n",
       "120          444           17       10                        1         200   \n",
       "121          444           18       10                        1         300   \n",
       "122          444           19       10                        1         300   \n",
       "123          444           20       10                        1         300   \n",
       "124          444           21       10                        1         300   \n",
       "\n",
       "     learning_rate  fitting_time  prediction_time  \\\n",
       "0         0.000046          3077              125   \n",
       "1         0.000089          3454              122   \n",
       "2         0.000443          2209              122   \n",
       "3         0.000753          1477              122   \n",
       "4         0.004071          1475              122   \n",
       "5         0.005730          1477              122   \n",
       "6         0.000027          1452              113   \n",
       "7         0.000094          1780              113   \n",
       "8         0.000462          1361              112   \n",
       "9         0.000946          1362              113   \n",
       "10        0.001557          1364              113   \n",
       "11        0.008522          1367              112   \n",
       "12        0.000021          1471              117   \n",
       "13        0.000090          1418              120   \n",
       "14        0.000157          1437              118   \n",
       "15        0.000587          1427              118   \n",
       "16        0.003289          1433              117   \n",
       "17        0.006520          1463              121   \n",
       "18        0.000022          1697              111   \n",
       "19        0.000095          1348              111   \n",
       "20        0.000065          1672               90   \n",
       "21        0.000065          1096               90   \n",
       "22        0.000428          1117               90   \n",
       "23        0.000428          1639               90   \n",
       "24        0.000770          1122               90   \n",
       "25        0.000770          1095               89   \n",
       "26        0.004850          1402               90   \n",
       "27        0.004850          1094               90   \n",
       "28        0.005693          1115               90   \n",
       "29        0.005693          1093               90   \n",
       "..             ...           ...              ...   \n",
       "95        0.000055          1485               88   \n",
       "96        0.000291          2319               88   \n",
       "97        0.000291          1580               88   \n",
       "98        0.000604          1632               88   \n",
       "99        0.000604          2604               88   \n",
       "100       0.001061          1603               88   \n",
       "101       0.001061          3042               88   \n",
       "102       0.000005          1640               88   \n",
       "103       0.000009          1457               96   \n",
       "104       0.000009          1494               95   \n",
       "105       0.000049          2015               95   \n",
       "106       0.000049          3112               97   \n",
       "107       0.000067          1361              107   \n",
       "108       0.000067          1356               96   \n",
       "109       0.000006          8158               96   \n",
       "110       0.000006          2091               96   \n",
       "111       0.000042          4922               96   \n",
       "112       0.000042          5297               96   \n",
       "113       0.000091          3212               96   \n",
       "114       0.000091          7417               96   \n",
       "115       0.000008          1739               98   \n",
       "116       0.000008          2345               99   \n",
       "117       0.000047          2393               99   \n",
       "118       0.000047          1686               98   \n",
       "119       0.000073          1624               98   \n",
       "120       0.000073          1815               98   \n",
       "121       0.000009          4357               98   \n",
       "122       0.000009          1672               99   \n",
       "123       0.000050          4380               99   \n",
       "124       0.000050          1811              135   \n",
       "\n",
       "                                         epochs_losses  \\\n",
       "0    [11.177390171632402, 5.857689602159346, 2.6551...   \n",
       "1    [10.441518563610224, 5.4293057353105985, 3.884...   \n",
       "2    [9.8417969667144, 5.32107679107467, 4.79954731...   \n",
       "3    [10.426884437054275, 7.350310457364854, 9.7848...   \n",
       "4    [14.32642020184331, 14.633708820949265, 14.633...   \n",
       "5    [14.377608573621332, 14.35784860733766, 14.357...   \n",
       "6    [12.859274773210181, 9.429001272356553, 6.2099...   \n",
       "7    [10.854263379801289, 5.383011704198885, 3.2654...   \n",
       "8    [9.503772049826363, 6.354559300223846, 6.12258...   \n",
       "9    [10.594130981094846, 9.908407009111276, 10.101...   \n",
       "10   [11.499933091153526, 10.687108548706917, 11.12...   \n",
       "11   [14.262176338438431, 14.409464812531487, 14.40...   \n",
       "12   [11.025774697125968, 7.274011296741033, 4.4514...   \n",
       "13   [10.217060638686357, 5.400305562100168, 3.1394...   \n",
       "14   [9.837554091114109, 3.447510222257194, 1.59543...   \n",
       "15   [10.605151224944551, 8.833039421146198, 8.9157...   \n",
       "16   [13.438748537483862, 13.53646965350135, 13.536...   \n",
       "17   [14.52452774371131, 14.56092522508007, 14.5609...   \n",
       "18   [12.375398138184083, 9.79439748513922, 5.40574...   \n",
       "19   [12.220591681222635, 7.596678482433391, 5.1738...   \n",
       "20   [9.61944356171981, 5.981166259102199, 5.565419...   \n",
       "21   [11.11389104179714, 6.314466725225034, 4.44965...   \n",
       "22   [13.487566305243451, 13.388840011928393, 14.05...   \n",
       "23   [7.219512410785841, 2.540557677331178, 1.89829...   \n",
       "24   [12.506660606550133, 13.595264476278555, 13.52...   \n",
       "25   [10.682513879693072, 6.306996697964876, 5.0812...   \n",
       "26   [13.795615154763926, 13.630303527997887, 13.56...   \n",
       "27   [13.70452408168627, 14.068295706873355, 14.068...   \n",
       "28   [13.924330690632695, 14.033256157584812, 14.03...   \n",
       "29   [13.995241020036781, 14.068295706873355, 14.06...   \n",
       "..                                                 ...   \n",
       "95   [12.531751771991173, 9.042676127358769, 5.7434...   \n",
       "96   [12.410123653626174, 9.598922086565683, 7.7693...   \n",
       "97   [11.956391655996944, 5.9083209198512385, 3.571...   \n",
       "98   [13.85870252030619, 13.854321597667223, 13.854...   \n",
       "99   [10.893494188115838, 5.097742683432076, 3.2496...   \n",
       "100  [13.505435204238035, 13.836211440268528, 13.83...   \n",
       "101  [10.081505084305666, 4.522153259663099, 3.4023...   \n",
       "102  [12.193586542365257, 10.661302737975388, 8.889...   \n",
       "103  [10.736687477159057, 7.303910214588492, 4.2432...   \n",
       "104  [12.933459277984651, 12.577445039070058, 11.97...   \n",
       "105  [8.19136002730536, 2.224062449422783, 0.796706...   \n",
       "106  [12.499718916059402, 9.08367666592908, 5.51035...   \n",
       "107  [7.601912935451827, 2.8883877705986407, 2.1372...   \n",
       "108  [13.238199061776585, 10.050782345396815, 7.420...   \n",
       "109  [12.094785657829544, 9.932752012837414, 7.9038...   \n",
       "110  [12.661890428501755, 12.380870878142838, 11.91...   \n",
       "111  [10.56009166528566, 4.163821252876022, 2.45548...   \n",
       "112  [12.84661759533011, 11.271038799463042, 9.4459...   \n",
       "113  [11.282455689397759, 8.86174060242833, 7.52606...   \n",
       "114  [12.795913973828956, 9.555369752110343, 4.9030...   \n",
       "115  [11.583652914836023, 8.805376688317896, 7.2445...   \n",
       "116  [12.272719182669212, 11.694719937500988, 10.68...   \n",
       "117  [8.738685776177263, 2.779545922612035, 1.40377...   \n",
       "118  [11.888713320779367, 8.879410602446152, 6.0340...   \n",
       "119  [11.166352626889251, 7.717718957528788, 6.3307...   \n",
       "120  [12.853760399157888, 9.65257440273875, 5.97255...   \n",
       "121  [12.051452930824581, 9.182866433030794, 6.9785...   \n",
       "122  [13.157143132872239, 12.729341613993245, 12.10...   \n",
       "123  [10.506150655486104, 5.842451920523803, 4.7496...   \n",
       "124  [13.170345185861308, 10.6054297804953, 7.32566...   \n",
       "\n",
       "                                     epochs_accuracies  \\\n",
       "0    [0.18174409032013714, 0.5069274632252613, 0.74...   \n",
       "1    [0.23553382233088835, 0.5973920026443407, 0.71...   \n",
       "2    [0.31621842245406606, 0.6479217584966739, 0.68...   \n",
       "3    [0.29910350730604723, 0.5264873691662211, 0.38...   \n",
       "4    [0.09942950298000179, 0.09209453961838526, 0.0...   \n",
       "5    [0.10268948660721697, 0.10920945234359913, 0.1...   \n",
       "6    [0.10424028090843042, 0.26855123606461095, 0.4...   \n",
       "7    [0.2049469942812785, 0.5927561940658219, 0.738...   \n",
       "8    [0.34098940186018234, 0.5697879856551072, 0.60...   \n",
       "9    [0.2985865757488004, 0.3780918724755938, 0.370...   \n",
       "10   [0.256183743160942, 0.3321554788747019, 0.3074...   \n",
       "11   [0.10689045948242973, 0.10600706758537057, 0.1...   \n",
       "12   [0.17796610106351013, 0.413559321629799, 0.605...   \n",
       "13   [0.25084745947082165, 0.6008474579301931, 0.77...   \n",
       "14   [0.28983050859454323, 0.7237288113367759, 0.86...   \n",
       "15   [0.31101694488424364, 0.44830507646172735, 0.4...   \n",
       "16   [0.15677966101694915, 0.16016949195477923, 0.1...   \n",
       "17   [0.09406779649651657, 0.09661016941575681, 0.0...   \n",
       "18   [0.12845673432688326, 0.2542372876304691, 0.52...   \n",
       "19   [0.17395182801983805, 0.4683318553919456, 0.62...   \n",
       "20   [0.35000000424359157, 0.6086956586526788, 0.64...   \n",
       "21   [0.18043478205800056, 0.5271739098040954, 0.67...   \n",
       "22   [0.14673913359318091, 0.16847826159842635, 0.1...   \n",
       "23   [0.43913043517133465, 0.8195652106533879, 0.85...   \n",
       "24   [0.21413043538189452, 0.15652173749454643, 0.1...   \n",
       "25   [0.29021738971705024, 0.5749999952057133, 0.67...   \n",
       "26   [0.14021739128815092, 0.15434782547147377, 0.1...   \n",
       "27   [0.13804347845523254, 0.12717391339981038, 0.1...   \n",
       "28   [0.12934782598977504, 0.12934782598977504, 0.1...   \n",
       "29   [0.12391304159941881, 0.1271739136427641, 0.12...   \n",
       "..                                                 ...   \n",
       "95   [0.10561798094363695, 0.3067415812042322, 0.52...   \n",
       "96   [0.19550561603535427, 0.3943820254186566, 0.51...   \n",
       "97   [0.15168539309099818, 0.5415730302253466, 0.74...   \n",
       "98   [0.12134831557782848, 0.14044943636053064, 0.1...   \n",
       "99   [0.22921348605932815, 0.6415730279483153, 0.76...   \n",
       "100  [0.14719100890869505, 0.14157303313860733, 0.1...   \n",
       "101  [0.2853932672002342, 0.6707865274354313, 0.773...   \n",
       "102  [0.12359550561797752, 0.19101123762934397, 0.2...   \n",
       "103  [0.2033023729966521, 0.4200206394043008, 0.629...   \n",
       "104  [0.09803921679348153, 0.12280701971214007, 0.1...   \n",
       "105  [0.3725490320946779, 0.8266253913027091, 0.933...   \n",
       "106  [0.10939112539384879, 0.297213626320025, 0.520...   \n",
       "107  [0.4334365318311134, 0.7843137323179727, 0.850...   \n",
       "108  [0.1073271394760124, 0.27863777055951955, 0.46...   \n",
       "109  [0.14447884455369353, 0.2662538664628847, 0.40...   \n",
       "110  [0.10010320072757202, 0.1207430337789258, 0.14...   \n",
       "111  [0.2569659387364107, 0.6718266078562191, 0.807...   \n",
       "112  [0.10423116778072558, 0.1991744041442871, 0.30...   \n",
       "113  [0.24871000315382755, 0.42208461689506155, 0.5...   \n",
       "114  [0.10216718238573694, 0.2745097998310538, 0.57...   \n",
       "115  [0.16986855689748837, 0.3488372009552646, 0.47...   \n",
       "116  [0.12841253913750664, 0.14459049290364384, 0.1...   \n",
       "117  [0.35389282259830207, 0.7724974700245023, 0.88...   \n",
       "118  [0.13751263845678047, 0.33063700689705366, 0.4...   \n",
       "119  [0.2709807908149291, 0.502527802007384, 0.5965...   \n",
       "120  [0.1213346834400908, 0.2942365967226945, 0.535...   \n",
       "121  [0.1678463114525358, 0.3053589442742968, 0.431...   \n",
       "122  [0.08695652133232422, 0.09302325837532599, 0.1...   \n",
       "123  [0.26289180596147194, 0.6036400478578072, 0.67...   \n",
       "124  [0.08392315420451131, 0.2082912012467611, 0.42...   \n",
       "\n",
       "                                     epochs_val_losses  \\\n",
       "0    [15.50204170040968, 15.00998678440001, 12.2796...   \n",
       "1    [15.783933965171256, 12.326112328506097, 12.78...   \n",
       "2    [12.235719762197355, 11.661158957132479, 11.52...   \n",
       "3    [16.118097258777155, 16.118097258777155, 16.11...   \n",
       "4    [16.118097258777155, 16.118097258777155, 16.11...   \n",
       "5    [16.118097258777155, 16.118097258777155, 16.11...   \n",
       "6    [15.520961927989173, 15.956519313590237, 16.03...   \n",
       "7    [16.11809730529785, 16.115518347926873, 16.118...   \n",
       "8    [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "9    [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "10   [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "11   [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "12   [15.697799256610386, 16.098241999669728, 16.09...   \n",
       "13   [16.111351148731213, 16.11809730529785, 16.118...   \n",
       "14   [16.117790386761506, 16.11809730529785, 16.118...   \n",
       "15   [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "16   [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "17   [16.11809730529785, 16.11809730529785, 16.1180...   \n",
       "18   [15.606216185870656, 15.482723521676293, 15.52...   \n",
       "19   [16.110363363582184, 16.11809730529785, 16.118...   \n",
       "20   [14.36760568774096, 15.284007802460016, 13.127...   \n",
       "21   [16.085105100749757, 16.107627346771935, 16.10...   \n",
       "22   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "23   [15.243229061462204, 13.201828611790162, 13.02...   \n",
       "24   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "25   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "26   [13.545501460469895, 13.387995778931858, 13.38...   \n",
       "27   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "28   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "29   [16.11809726180782, 16.11809726180782, 16.1180...   \n",
       "..                                                 ...   \n",
       "95   [15.782424457948215, 16.034975787204523, 16.10...   \n",
       "96   [14.268121334037396, 14.528891213413843, 13.96...   \n",
       "97   [13.86413397612395, 14.419594356909345, 14.405...   \n",
       "98   [16.118099874117558, 16.118099874117558, 16.11...   \n",
       "99   [14.632227765992033, 13.788364712236707, 14.00...   \n",
       "100  [16.118099874117558, 16.118099874117558, 16.11...   \n",
       "101  [14.218336965901281, 13.99168165444525, 13.990...   \n",
       "102  [14.465690612792969, 14.751871109008789, 14.88...   \n",
       "103  [15.823758878205952, 16.04527726778674, 16.097...   \n",
       "104  [14.019623103894686, 14.42137057714787, 14.778...   \n",
       "105  [13.19497582639334, 12.653280547528814, 12.879...   \n",
       "106  [14.741868503322543, 15.533892785063468, 14.81...   \n",
       "107  [16.11810039366731, 16.11810039366731, 16.1181...   \n",
       "108  [15.46262528800374, 16.09364617645925, 16.1085...   \n",
       "109  [15.704390956890473, 15.926995829532022, 15.79...   \n",
       "110  [15.218670963133821, 15.288978062916097, 15.38...   \n",
       "111  [13.896825524675588, 12.75900555542748, 12.684...   \n",
       "112  [13.369037610458516, 14.198510066643587, 13.98...   \n",
       "113  [12.769200717701631, 13.259642054791051, 12.75...   \n",
       "114  [14.268447320896774, 13.792559396371752, 13.49...   \n",
       "115  [15.894092126326127, 16.023284276326496, 16.04...   \n",
       "116  [15.269959218574293, 15.42425502430309, 15.624...   \n",
       "117  [16.063047351259172, 15.81689707438151, 12.779...   \n",
       "118  [15.033767873590643, 16.032680569273055, 16.10...   \n",
       "119  [16.11810036861535, 16.11810036861535, 16.1181...   \n",
       "120  [15.856012170965021, 16.108810771595348, 16.11...   \n",
       "121  [14.675517342307351, 14.236758665605025, 14.17...   \n",
       "122  [13.869226455688477, 14.43110214580189, 15.068...   \n",
       "123  [16.118100599809125, 16.118100599809125, 16.05...   \n",
       "124  [14.294045188210227, 15.977744015780361, 16.10...   \n",
       "\n",
       "                                 epochs_val_accuracies  test_loss  \\\n",
       "0    [0.0, 0.029268292483033205, 0.2048780423838918...   3.816269   \n",
       "1    [0.002439024335727459, 0.19268292479398774, 0....   3.838663   \n",
       "2    [0.21219512674866653, 0.2682926887419166, 0.27...   7.138068   \n",
       "3                                 [0.0, 0.0, 0.0, 0.0]  10.832995   \n",
       "4                                 [0.0, 0.0, 0.0, 0.0]  15.008516   \n",
       "5                                 [0.0, 0.0, 0.0, 0.0]  14.804120   \n",
       "6    [0.00793650775911316, 0.0, 0.0, 0.002645502586...   6.095660   \n",
       "7                            [0.0, 0.0, 0.0, 0.0, 0.0]   5.329826   \n",
       "8                                 [0.0, 0.0, 0.0, 0.0]   8.239412   \n",
       "9                                 [0.0, 0.0, 0.0, 0.0]  12.452109   \n",
       "10                                [0.0, 0.0, 0.0, 0.0]  12.548790   \n",
       "11                                [0.0, 0.0, 0.0, 0.0]  14.822327   \n",
       "12               [0.007614213027778616, 0.0, 0.0, 0.0]   5.951484   \n",
       "13                                [0.0, 0.0, 0.0, 0.0]   6.159383   \n",
       "14                                [0.0, 0.0, 0.0, 0.0]   5.221603   \n",
       "15                                [0.0, 0.0, 0.0, 0.0]  10.933599   \n",
       "16                                [0.0, 0.0, 0.0, 0.0]  14.202171   \n",
       "17                                [0.0, 0.0, 0.0, 0.0]  14.932047   \n",
       "18   [0.013368984156432636, 0.029411764546511645, 0...   5.076590   \n",
       "19                                [0.0, 0.0, 0.0, 0.0]   6.485281   \n",
       "20   [0.09120521211468824, 0.04560260605734412, 0.1...   7.928827   \n",
       "21                                [0.0, 0.0, 0.0, 0.0]   6.912676   \n",
       "22                                [0.0, 0.0, 0.0, 0.0]  14.560792   \n",
       "23   [0.026058631339368293, 0.17915309834557946, 0....   5.245852   \n",
       "24                                [0.0, 0.0, 0.0, 0.0]  14.171466   \n",
       "25                                [0.0, 0.0, 0.0, 0.0]   9.279782   \n",
       "26   [0.15960912362760363, 0.1693811012789947, 0.16...  13.159218   \n",
       "27                                [0.0, 0.0, 0.0, 0.0]  14.560792   \n",
       "28                                [0.0, 0.0, 0.0, 0.0]  14.560792   \n",
       "29                                [0.0, 0.0, 0.0, 0.0]  14.560792   \n",
       "..                                                 ...        ...   \n",
       "95               [0.003367003263523081, 0.0, 0.0, 0.0]   5.981250   \n",
       "96   [0.11111111552626998, 0.09427609467747236, 0.1...  10.516670   \n",
       "97   [0.12794612634061564, 0.08080807900187945, 0.1...   5.823327   \n",
       "98                                [0.0, 0.0, 0.0, 0.0]  14.425695   \n",
       "99   [0.05387205266791963, 0.14141413699898253, 0.1...   4.464179   \n",
       "100                               [0.0, 0.0, 0.0, 0.0]  14.425695   \n",
       "101  [0.1010101050238818, 0.1245791261846369, 0.131...   5.603336   \n",
       "102  [0.033670034259557724, 0.04377104341983795, 0....   8.045590   \n",
       "103                               [0.0, 0.0, 0.0, 0.0]   5.366906   \n",
       "104  [0.024767801834518325, 0.012383900605857187, 0...  10.915709   \n",
       "105  [0.14551083554424368, 0.21052631800388774, 0.1...   3.637250   \n",
       "106  [0.012383900651990814, 0.006191950325995407, 0...   3.831526   \n",
       "107                               [0.0, 0.0, 0.0, 0.0]   5.459007   \n",
       "108              [0.006191950325995407, 0.0, 0.0, 0.0]   8.237578   \n",
       "109  [0.0, 0.0, 0.0, 0.0030959753071652845, 0.04643...   5.026830   \n",
       "110              [0.0, 0.0, 0.0, 0.006191950614330569]  11.176276   \n",
       "111  [0.086687306870617, 0.19195046125919826, 0.204...   3.594073   \n",
       "112  [0.05572755293395866, 0.03715170195597244, 0.0...   6.137263   \n",
       "113  [0.1981424196585782, 0.16099071078256189, 0.20...   6.862218   \n",
       "114  [0.0433436534353085, 0.08359133459097091, 0.12...   3.549291   \n",
       "115                               [0.0, 0.0, 0.0, 0.0]   7.677936   \n",
       "116  [0.012121211850281918, 0.0030303029625704794, ...   9.369028   \n",
       "117  [0.0, 0.006060605925140959, 0.1939393896045106...   3.732816   \n",
       "118                [0.0121212119913914, 0.0, 0.0, 0.0]   5.818482   \n",
       "119                               [0.0, 0.0, 0.0, 0.0]   8.026507   \n",
       "120               [0.00303030303312522, 0.0, 0.0, 0.0]   6.849744   \n",
       "121  [0.03333333202383735, 0.06060606376691298, 0.0...   3.974939   \n",
       "122  [0.048484849658879364, 0.036363635550845756, 0...  11.600196   \n",
       "123  [0.0, 0.0, 0.0030303031036799603, 0.0, 0.00909...   5.505957   \n",
       "124              [0.009090908887711439, 0.0, 0.0, 0.0]   6.114878   \n",
       "\n",
       "     test_accuracy  is_augmented optimizer learning_rate_categ    run  \n",
       "0         0.722826         False       sgd         [1e-5,5e-5]  run_1  \n",
       "1         0.733696         False       sgd         [5e-5,1e-4]  run_1  \n",
       "2         0.543478         False       sgd         [1e-4,5e-4]  run_1  \n",
       "3         0.327899         False       sgd         [5e-4,1e-3]  run_1  \n",
       "4         0.068841         False       sgd         [1e-3,5e-3]  run_1  \n",
       "5         0.081522         False       sgd         [5e-3,1e-2]  run_1  \n",
       "6         0.547059         False       sgd         [1e-5,5e-5]  run_1  \n",
       "7         0.633333         False       sgd         [5e-5,1e-4]  run_1  \n",
       "8         0.484314         False       sgd         [1e-4,5e-4]  run_1  \n",
       "9         0.227451         False       sgd         [5e-4,1e-3]  run_1  \n",
       "10        0.219608         False       sgd         [1e-3,5e-3]  run_1  \n",
       "11        0.080392         False       sgd         [5e-3,1e-2]  run_1  \n",
       "12        0.566038         False       sgd         [1e-5,5e-5]  run_1  \n",
       "13        0.592453         False       sgd         [5e-5,1e-4]  run_1  \n",
       "14        0.652830         False       sgd         [1e-4,5e-4]  run_1  \n",
       "15        0.316981         False       sgd         [5e-4,1e-3]  run_1  \n",
       "16        0.118868         False       sgd         [1e-3,5e-3]  run_1  \n",
       "17        0.073585         False       sgd         [5e-3,1e-2]  run_1  \n",
       "18        0.619802         False       sgd         [1e-5,5e-5]  run_1  \n",
       "19        0.562376         False       sgd         [5e-5,1e-4]  run_1  \n",
       "20        0.497585         False      adam         [5e-5,1e-4]  run_1  \n",
       "21        0.533816         False       sgd         [5e-5,1e-4]  run_1  \n",
       "22        0.096618         False      adam         [1e-4,5e-4]  run_1  \n",
       "23        0.659420         False       sgd         [1e-4,5e-4]  run_1  \n",
       "24        0.120773         False      adam         [5e-4,1e-3]  run_1  \n",
       "25        0.420290         False       sgd         [5e-4,1e-3]  run_1  \n",
       "26        0.183575         False      adam         [1e-3,5e-3]  run_1  \n",
       "27        0.096618         False       sgd         [1e-3,5e-3]  run_1  \n",
       "28        0.096618         False      adam         [5e-3,1e-2]  run_1  \n",
       "29        0.096618         False       sgd         [5e-3,1e-2]  run_1  \n",
       "..             ...           ...       ...                 ...    ...  \n",
       "95        0.570000         False       sgd         [5e-5,1e-4]  run_1  \n",
       "96        0.342500         False      adam         [1e-4,5e-4]  run_1  \n",
       "97        0.607500         False       sgd         [1e-4,5e-4]  run_1  \n",
       "98        0.105000         False      adam         [5e-4,1e-3]  run_1  \n",
       "99        0.710000         False       sgd         [5e-4,1e-3]  run_1  \n",
       "100       0.105000         False      adam         [1e-3,5e-3]  run_1  \n",
       "101       0.640000         False       sgd         [1e-3,5e-3]  run_1  \n",
       "102       0.377500         False      adam         [5e-6,1e-5]  run_1  \n",
       "103       0.642032         False      adam                 NaN  run_2  \n",
       "104       0.193995         False       sgd                 NaN  run_2  \n",
       "105       0.764434         False      adam                 NaN  run_2  \n",
       "106       0.732102         False       sgd                 NaN  run_2  \n",
       "107       0.655889         False      adam                 NaN  run_2  \n",
       "108       0.475751         False       sgd                 NaN  run_2  \n",
       "109       0.644342         False      adam                 NaN  run_2  \n",
       "110       0.168591         False       sgd                 NaN  run_2  \n",
       "111       0.764434         False      adam                 NaN  run_2  \n",
       "112       0.554273         False       sgd                 NaN  run_2  \n",
       "113       0.568129         False      adam                 NaN  run_2  \n",
       "114       0.764434         False       sgd                 NaN  run_2  \n",
       "115       0.452703         False      adam                 NaN  run_2  \n",
       "116       0.234234         False       sgd                 NaN  run_2  \n",
       "117       0.750000         False      adam                 NaN  run_2  \n",
       "118       0.578829         False       sgd                 NaN  run_2  \n",
       "119       0.500000         False      adam                 NaN  run_2  \n",
       "120       0.524775         False       sgd                 NaN  run_2  \n",
       "121       0.707207         False      adam                 NaN  run_2  \n",
       "122       0.150901         False       sgd                 NaN  run_2  \n",
       "123       0.650901         False      adam                 NaN  run_2  \n",
       "124       0.529279         False       sgd                 NaN  run_2  \n",
       "\n",
       "[125 rows x 20 columns]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation = pd.read_csv('df_evaluation_learning_rates.csv')\n",
    "df_evaluation.drop('Unnamed: 0', axis=1, inplace=True)\n",
    "df_evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['races', 'training_len', 'testing_len', 'iter_number', 'n_races',\n",
       "       'random_selection_number', 'batch_size', 'learning_rate',\n",
       "       'fitting_time', 'prediction_time', 'epochs_losses', 'epochs_accuracies',\n",
       "       'epochs_val_losses', 'epochs_val_accuracies', 'test_loss',\n",
       "       'test_accuracy', 'is_augmented', 'optimizer', 'learning_rate_categ',\n",
       "       'run', 'n_epochs', 'final_loss', 'final_accuracy', 'final_val_loss',\n",
       "       'final_val_accuracy'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">I) Preprocessing over the results</font>**\n",
    "</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7644341801385681"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.max(df_evaluation['test_accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "for column in ['epochs_losses','epochs_accuracies','epochs_val_losses','epochs_val_accuracies']:\n",
    "    df_evaluation = convert_in_list(df_evaluation,column)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[11.177390171632402,\n",
       " 5.857689602159346,\n",
       " 2.6551777008790354,\n",
       " 1.5934128820920437,\n",
       " 1.1124129741290962,\n",
       " 0.7947726992202563,\n",
       " 0.7803367870361123,\n",
       " 0.5023115000900019]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation['epochs_losses'][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_evaluation['n_epochs'] = df_evaluation['epochs_losses'].apply(lambda x: len(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>races</th>\n",
       "      <th>training_len</th>\n",
       "      <th>testing_len</th>\n",
       "      <th>iter_number</th>\n",
       "      <th>n_races</th>\n",
       "      <th>random_selection_number</th>\n",
       "      <th>batch_size</th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>fitting_time</th>\n",
       "      <th>prediction_time</th>\n",
       "      <th>...</th>\n",
       "      <th>test_accuracy</th>\n",
       "      <th>is_augmented</th>\n",
       "      <th>optimizer</th>\n",
       "      <th>learning_rate_categ</th>\n",
       "      <th>run</th>\n",
       "      <th>n_epochs</th>\n",
       "      <th>final_loss</th>\n",
       "      <th>final_accuracy</th>\n",
       "      <th>final_val_loss</th>\n",
       "      <th>final_val_accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>3077</td>\n",
       "      <td>125</td>\n",
       "      <td>...</td>\n",
       "      <td>0.722826</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>8</td>\n",
       "      <td>0.502312</td>\n",
       "      <td>0.947025</td>\n",
       "      <td>11.928421</td>\n",
       "      <td>0.239024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000089</td>\n",
       "      <td>3454</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.733696</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>9</td>\n",
       "      <td>0.311288</td>\n",
       "      <td>0.967400</td>\n",
       "      <td>11.619101</td>\n",
       "      <td>0.263415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>2</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000443</td>\n",
       "      <td>2209</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.543478</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>4.411082</td>\n",
       "      <td>0.719641</td>\n",
       "      <td>12.300757</td>\n",
       "      <td>0.221951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>3</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000753</td>\n",
       "      <td>1477</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.327899</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>9.715977</td>\n",
       "      <td>0.392013</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>4</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004071</td>\n",
       "      <td>1475</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.068841</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.633709</td>\n",
       "      <td>0.092095</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>5</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005730</td>\n",
       "      <td>1477</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.081522</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.357848</td>\n",
       "      <td>0.109209</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>6</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000027</td>\n",
       "      <td>1452</td>\n",
       "      <td>113</td>\n",
       "      <td>...</td>\n",
       "      <td>0.547059</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>4.070631</td>\n",
       "      <td>0.652827</td>\n",
       "      <td>16.028574</td>\n",
       "      <td>0.002646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>7</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000094</td>\n",
       "      <td>1780</td>\n",
       "      <td>113</td>\n",
       "      <td>...</td>\n",
       "      <td>0.633333</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>5</td>\n",
       "      <td>1.477193</td>\n",
       "      <td>0.879859</td>\n",
       "      <td>16.117563</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>8</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000462</td>\n",
       "      <td>1361</td>\n",
       "      <td>112</td>\n",
       "      <td>...</td>\n",
       "      <td>0.484314</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>5.677927</td>\n",
       "      <td>0.639576</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>9</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000946</td>\n",
       "      <td>1362</td>\n",
       "      <td>113</td>\n",
       "      <td>...</td>\n",
       "      <td>0.227451</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>10.009441</td>\n",
       "      <td>0.376325</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>10</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.001557</td>\n",
       "      <td>1364</td>\n",
       "      <td>113</td>\n",
       "      <td>...</td>\n",
       "      <td>0.219608</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>11.638678</td>\n",
       "      <td>0.275618</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>['brabancon_griffon', 'cairn', 'cocker_spaniel...</td>\n",
       "      <td>1510</td>\n",
       "      <td>510</td>\n",
       "      <td>11</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.008522</td>\n",
       "      <td>1367</td>\n",
       "      <td>112</td>\n",
       "      <td>...</td>\n",
       "      <td>0.080392</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.409465</td>\n",
       "      <td>0.106007</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>12</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000021</td>\n",
       "      <td>1471</td>\n",
       "      <td>117</td>\n",
       "      <td>...</td>\n",
       "      <td>0.566038</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>2.949758</td>\n",
       "      <td>0.725424</td>\n",
       "      <td>16.100197</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>13</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000090</td>\n",
       "      <td>1418</td>\n",
       "      <td>120</td>\n",
       "      <td>...</td>\n",
       "      <td>0.592453</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>2.432044</td>\n",
       "      <td>0.826271</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>14</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000157</td>\n",
       "      <td>1437</td>\n",
       "      <td>118</td>\n",
       "      <td>...</td>\n",
       "      <td>0.652830</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>1.248966</td>\n",
       "      <td>0.896610</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>15</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000587</td>\n",
       "      <td>1427</td>\n",
       "      <td>118</td>\n",
       "      <td>...</td>\n",
       "      <td>0.316981</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>8.976806</td>\n",
       "      <td>0.438983</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>16</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.003289</td>\n",
       "      <td>1433</td>\n",
       "      <td>117</td>\n",
       "      <td>...</td>\n",
       "      <td>0.118868</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>13.536469</td>\n",
       "      <td>0.160169</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>['beagle', 'bloodhound', 'border_collie', 'bou...</td>\n",
       "      <td>1574</td>\n",
       "      <td>530</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.006520</td>\n",
       "      <td>1463</td>\n",
       "      <td>121</td>\n",
       "      <td>...</td>\n",
       "      <td>0.073585</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.560925</td>\n",
       "      <td>0.096610</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>['bluetick', 'border_collie', 'border_terrier'...</td>\n",
       "      <td>1495</td>\n",
       "      <td>505</td>\n",
       "      <td>18</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000022</td>\n",
       "      <td>1697</td>\n",
       "      <td>111</td>\n",
       "      <td>...</td>\n",
       "      <td>0.619802</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>5</td>\n",
       "      <td>2.081406</td>\n",
       "      <td>0.797502</td>\n",
       "      <td>15.511462</td>\n",
       "      <td>0.034759</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>['bluetick', 'border_collie', 'border_terrier'...</td>\n",
       "      <td>1495</td>\n",
       "      <td>505</td>\n",
       "      <td>19</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000095</td>\n",
       "      <td>1348</td>\n",
       "      <td>111</td>\n",
       "      <td>...</td>\n",
       "      <td>0.562376</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>3.698544</td>\n",
       "      <td>0.727922</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1672</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.497585</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>5.101280</td>\n",
       "      <td>0.679348</td>\n",
       "      <td>13.880275</td>\n",
       "      <td>0.133550</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1096</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.533816</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>4.101719</td>\n",
       "      <td>0.709783</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000428</td>\n",
       "      <td>1117</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.068295</td>\n",
       "      <td>0.127174</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000428</td>\n",
       "      <td>1639</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.659420</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>1.035343</td>\n",
       "      <td>0.926087</td>\n",
       "      <td>13.074652</td>\n",
       "      <td>0.188925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000770</td>\n",
       "      <td>1122</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.120773</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>13.542705</td>\n",
       "      <td>0.159783</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000770</td>\n",
       "      <td>1095</td>\n",
       "      <td>89</td>\n",
       "      <td>...</td>\n",
       "      <td>0.420290</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>5.632835</td>\n",
       "      <td>0.641304</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>1402</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.183575</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>5</td>\n",
       "      <td>13.490146</td>\n",
       "      <td>0.163043</td>\n",
       "      <td>13.387996</td>\n",
       "      <td>0.169381</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>1094</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.068295</td>\n",
       "      <td>0.127174</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005693</td>\n",
       "      <td>1115</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.033256</td>\n",
       "      <td>0.129348</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.005693</td>\n",
       "      <td>1093</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.096618</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-3,1e-2]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>14.068296</td>\n",
       "      <td>0.127174</td>\n",
       "      <td>16.118097</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000055</td>\n",
       "      <td>1485</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.570000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>3.775522</td>\n",
       "      <td>0.680899</td>\n",
       "      <td>16.103250</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000291</td>\n",
       "      <td>2319</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.342500</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>8.938577</td>\n",
       "      <td>0.442697</td>\n",
       "      <td>14.652818</td>\n",
       "      <td>0.090909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000291</td>\n",
       "      <td>1580</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.607500</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>2.770390</td>\n",
       "      <td>0.802247</td>\n",
       "      <td>13.964347</td>\n",
       "      <td>0.127946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000604</td>\n",
       "      <td>1632</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.105000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>13.854322</td>\n",
       "      <td>0.140449</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000604</td>\n",
       "      <td>2604</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.710000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>7</td>\n",
       "      <td>0.777026</td>\n",
       "      <td>0.938202</td>\n",
       "      <td>14.086784</td>\n",
       "      <td>0.124579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>1603</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.105000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>13.836212</td>\n",
       "      <td>0.141573</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>3042</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.640000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>9</td>\n",
       "      <td>2.322723</td>\n",
       "      <td>0.852809</td>\n",
       "      <td>13.784502</td>\n",
       "      <td>0.144781</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>1640</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.377500</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-6,1e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>7.517145</td>\n",
       "      <td>0.361798</td>\n",
       "      <td>14.940201</td>\n",
       "      <td>0.033670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1457</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.642032</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>2.946247</td>\n",
       "      <td>0.735810</td>\n",
       "      <td>16.089482</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>104</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1494</td>\n",
       "      <td>95</td>\n",
       "      <td>...</td>\n",
       "      <td>0.193995</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>10.925501</td>\n",
       "      <td>0.202270</td>\n",
       "      <td>15.011213</td>\n",
       "      <td>0.012384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>2015</td>\n",
       "      <td>95</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>5</td>\n",
       "      <td>0.295404</td>\n",
       "      <td>0.974200</td>\n",
       "      <td>12.694006</td>\n",
       "      <td>0.204334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>3112</td>\n",
       "      <td>97</td>\n",
       "      <td>...</td>\n",
       "      <td>0.732102</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>8</td>\n",
       "      <td>0.475986</td>\n",
       "      <td>0.942208</td>\n",
       "      <td>13.152557</td>\n",
       "      <td>0.173375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>107</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000067</td>\n",
       "      <td>1361</td>\n",
       "      <td>107</td>\n",
       "      <td>...</td>\n",
       "      <td>0.655889</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>1.855424</td>\n",
       "      <td>0.875129</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>108</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>5</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000067</td>\n",
       "      <td>1356</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.475751</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>6.235126</td>\n",
       "      <td>0.574819</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>8158</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.644342</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>15</td>\n",
       "      <td>2.279800</td>\n",
       "      <td>0.829721</td>\n",
       "      <td>12.834111</td>\n",
       "      <td>0.195046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>2091</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.168591</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>11.222563</td>\n",
       "      <td>0.175439</td>\n",
       "      <td>15.410839</td>\n",
       "      <td>0.006192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>4922</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>9</td>\n",
       "      <td>0.230859</td>\n",
       "      <td>0.981424</td>\n",
       "      <td>12.996278</td>\n",
       "      <td>0.188854</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>5297</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.554273</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>3.820022</td>\n",
       "      <td>0.703818</td>\n",
       "      <td>13.087998</td>\n",
       "      <td>0.160991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>3212</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.568129</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>6</td>\n",
       "      <td>5.352686</td>\n",
       "      <td>0.652219</td>\n",
       "      <td>13.468336</td>\n",
       "      <td>0.154799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>7417</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>14</td>\n",
       "      <td>0.125814</td>\n",
       "      <td>0.984520</td>\n",
       "      <td>13.115022</td>\n",
       "      <td>0.173375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000008</td>\n",
       "      <td>1739</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.452703</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>6.168376</td>\n",
       "      <td>0.548028</td>\n",
       "      <td>16.026630</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>13</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000008</td>\n",
       "      <td>2345</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.234234</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>9.926596</td>\n",
       "      <td>0.230536</td>\n",
       "      <td>15.771304</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>14</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>2393</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>6</td>\n",
       "      <td>0.403706</td>\n",
       "      <td>0.970677</td>\n",
       "      <td>12.921979</td>\n",
       "      <td>0.178788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>118</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>15</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>1686</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.578829</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>3.807181</td>\n",
       "      <td>0.656218</td>\n",
       "      <td>16.117965</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>16</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000073</td>\n",
       "      <td>1624</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>5.519053</td>\n",
       "      <td>0.648129</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>17</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000073</td>\n",
       "      <td>1815</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.524775</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>4.053692</td>\n",
       "      <td>0.683519</td>\n",
       "      <td>16.118100</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>18</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>4357</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.707207</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>0.552625</td>\n",
       "      <td>0.931244</td>\n",
       "      <td>13.424177</td>\n",
       "      <td>0.142424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>19</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1672</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.150901</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>11.656408</td>\n",
       "      <td>0.140546</td>\n",
       "      <td>15.454955</td>\n",
       "      <td>0.012121</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>4380</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.650901</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>2.354338</td>\n",
       "      <td>0.852376</td>\n",
       "      <td>13.006859</td>\n",
       "      <td>0.181818</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>21</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>1811</td>\n",
       "      <td>135</td>\n",
       "      <td>...</td>\n",
       "      <td>0.529279</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>run_2</td>\n",
       "      <td>4</td>\n",
       "      <td>4.938865</td>\n",
       "      <td>0.587462</td>\n",
       "      <td>16.111250</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>125 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 races  training_len  \\\n",
       "0    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "1    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "2    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "3    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "4    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "5    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "6    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "7    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "8    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "9    ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "10   ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "11   ['brabancon_griffon', 'cairn', 'cocker_spaniel...          1510   \n",
       "12   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "13   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "14   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "15   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "16   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "17   ['beagle', 'bloodhound', 'border_collie', 'bou...          1574   \n",
       "18   ['bluetick', 'border_collie', 'border_terrier'...          1495   \n",
       "19   ['bluetick', 'border_collie', 'border_terrier'...          1495   \n",
       "20   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "21   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "22   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "23   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "24   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "25   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "26   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "27   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "28   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "29   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "..                                                 ...           ...   \n",
       "95   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "96   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "97   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "98   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "99   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "100  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "101  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "102  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "103  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "104  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "105  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "106  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "107  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "108  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "109  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "110  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "111  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "112  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "113  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "114  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "115  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "116  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "117  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "118  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "119  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "120  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "121  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "122  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "123  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "124  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "\n",
       "     testing_len  iter_number  n_races  random_selection_number  batch_size  \\\n",
       "0            552            0       12                        0         100   \n",
       "1            552            1       12                        0         100   \n",
       "2            552            2       12                        0         100   \n",
       "3            552            3       12                        0         100   \n",
       "4            552            4       12                        0         100   \n",
       "5            552            5       12                        0         100   \n",
       "6            510            6       12                        1         100   \n",
       "7            510            7       12                        1         100   \n",
       "8            510            8       12                        1         100   \n",
       "9            510            9       12                        1         100   \n",
       "10           510           10       12                        1         100   \n",
       "11           510           11       12                        1         100   \n",
       "12           530           12       12                        2         100   \n",
       "13           530           13       12                        2         100   \n",
       "14           530           14       12                        2         100   \n",
       "15           530           15       12                        2         100   \n",
       "16           530           16       12                        2         100   \n",
       "17           530           17       12                        2         100   \n",
       "18           505           18       12                        3         100   \n",
       "19           505           19       12                        3         100   \n",
       "20           414            0       10                        0         100   \n",
       "21           414            1       10                        0         100   \n",
       "22           414            2       10                        0         100   \n",
       "23           414            3       10                        0         100   \n",
       "24           414            4       10                        0         100   \n",
       "25           414            5       10                        0         100   \n",
       "26           414            6       10                        0         100   \n",
       "27           414            7       10                        0         100   \n",
       "28           414            8       10                        0         100   \n",
       "29           414            9       10                        0         100   \n",
       "..           ...          ...      ...                      ...         ...   \n",
       "95           400            5       10                        0         200   \n",
       "96           400            6       10                        0         200   \n",
       "97           400            7       10                        0         200   \n",
       "98           400            8       10                        0         200   \n",
       "99           400            9       10                        0         200   \n",
       "100          400           10       10                        0         200   \n",
       "101          400           11       10                        0         200   \n",
       "102          400           12       10                        0         300   \n",
       "103          433            0       10                        0         200   \n",
       "104          433            1       10                        0         200   \n",
       "105          433            2       10                        0         200   \n",
       "106          433            3       10                        0         200   \n",
       "107          433            4       10                        0         200   \n",
       "108          433            5       10                        0         200   \n",
       "109          433            6       10                        0         300   \n",
       "110          433            7       10                        0         300   \n",
       "111          433            8       10                        0         300   \n",
       "112          433            9       10                        0         300   \n",
       "113          433           10       10                        0         300   \n",
       "114          433           11       10                        0         300   \n",
       "115          444           12       10                        1         200   \n",
       "116          444           13       10                        1         200   \n",
       "117          444           14       10                        1         200   \n",
       "118          444           15       10                        1         200   \n",
       "119          444           16       10                        1         200   \n",
       "120          444           17       10                        1         200   \n",
       "121          444           18       10                        1         300   \n",
       "122          444           19       10                        1         300   \n",
       "123          444           20       10                        1         300   \n",
       "124          444           21       10                        1         300   \n",
       "\n",
       "     learning_rate  fitting_time  prediction_time        ...          \\\n",
       "0         0.000046          3077              125        ...           \n",
       "1         0.000089          3454              122        ...           \n",
       "2         0.000443          2209              122        ...           \n",
       "3         0.000753          1477              122        ...           \n",
       "4         0.004071          1475              122        ...           \n",
       "5         0.005730          1477              122        ...           \n",
       "6         0.000027          1452              113        ...           \n",
       "7         0.000094          1780              113        ...           \n",
       "8         0.000462          1361              112        ...           \n",
       "9         0.000946          1362              113        ...           \n",
       "10        0.001557          1364              113        ...           \n",
       "11        0.008522          1367              112        ...           \n",
       "12        0.000021          1471              117        ...           \n",
       "13        0.000090          1418              120        ...           \n",
       "14        0.000157          1437              118        ...           \n",
       "15        0.000587          1427              118        ...           \n",
       "16        0.003289          1433              117        ...           \n",
       "17        0.006520          1463              121        ...           \n",
       "18        0.000022          1697              111        ...           \n",
       "19        0.000095          1348              111        ...           \n",
       "20        0.000065          1672               90        ...           \n",
       "21        0.000065          1096               90        ...           \n",
       "22        0.000428          1117               90        ...           \n",
       "23        0.000428          1639               90        ...           \n",
       "24        0.000770          1122               90        ...           \n",
       "25        0.000770          1095               89        ...           \n",
       "26        0.004850          1402               90        ...           \n",
       "27        0.004850          1094               90        ...           \n",
       "28        0.005693          1115               90        ...           \n",
       "29        0.005693          1093               90        ...           \n",
       "..             ...           ...              ...        ...           \n",
       "95        0.000055          1485               88        ...           \n",
       "96        0.000291          2319               88        ...           \n",
       "97        0.000291          1580               88        ...           \n",
       "98        0.000604          1632               88        ...           \n",
       "99        0.000604          2604               88        ...           \n",
       "100       0.001061          1603               88        ...           \n",
       "101       0.001061          3042               88        ...           \n",
       "102       0.000005          1640               88        ...           \n",
       "103       0.000009          1457               96        ...           \n",
       "104       0.000009          1494               95        ...           \n",
       "105       0.000049          2015               95        ...           \n",
       "106       0.000049          3112               97        ...           \n",
       "107       0.000067          1361              107        ...           \n",
       "108       0.000067          1356               96        ...           \n",
       "109       0.000006          8158               96        ...           \n",
       "110       0.000006          2091               96        ...           \n",
       "111       0.000042          4922               96        ...           \n",
       "112       0.000042          5297               96        ...           \n",
       "113       0.000091          3212               96        ...           \n",
       "114       0.000091          7417               96        ...           \n",
       "115       0.000008          1739               98        ...           \n",
       "116       0.000008          2345               99        ...           \n",
       "117       0.000047          2393               99        ...           \n",
       "118       0.000047          1686               98        ...           \n",
       "119       0.000073          1624               98        ...           \n",
       "120       0.000073          1815               98        ...           \n",
       "121       0.000009          4357               98        ...           \n",
       "122       0.000009          1672               99        ...           \n",
       "123       0.000050          4380               99        ...           \n",
       "124       0.000050          1811              135        ...           \n",
       "\n",
       "    test_accuracy is_augmented optimizer learning_rate_categ    run  n_epochs  \\\n",
       "0        0.722826        False       sgd         [1e-5,5e-5]  run_1         8   \n",
       "1        0.733696        False       sgd         [5e-5,1e-4]  run_1         9   \n",
       "2        0.543478        False       sgd         [1e-4,5e-4]  run_1         6   \n",
       "3        0.327899        False       sgd         [5e-4,1e-3]  run_1         4   \n",
       "4        0.068841        False       sgd         [1e-3,5e-3]  run_1         4   \n",
       "5        0.081522        False       sgd         [5e-3,1e-2]  run_1         4   \n",
       "6        0.547059        False       sgd         [1e-5,5e-5]  run_1         4   \n",
       "7        0.633333        False       sgd         [5e-5,1e-4]  run_1         5   \n",
       "8        0.484314        False       sgd         [1e-4,5e-4]  run_1         4   \n",
       "9        0.227451        False       sgd         [5e-4,1e-3]  run_1         4   \n",
       "10       0.219608        False       sgd         [1e-3,5e-3]  run_1         4   \n",
       "11       0.080392        False       sgd         [5e-3,1e-2]  run_1         4   \n",
       "12       0.566038        False       sgd         [1e-5,5e-5]  run_1         4   \n",
       "13       0.592453        False       sgd         [5e-5,1e-4]  run_1         4   \n",
       "14       0.652830        False       sgd         [1e-4,5e-4]  run_1         4   \n",
       "15       0.316981        False       sgd         [5e-4,1e-3]  run_1         4   \n",
       "16       0.118868        False       sgd         [1e-3,5e-3]  run_1         4   \n",
       "17       0.073585        False       sgd         [5e-3,1e-2]  run_1         4   \n",
       "18       0.619802        False       sgd         [1e-5,5e-5]  run_1         5   \n",
       "19       0.562376        False       sgd         [5e-5,1e-4]  run_1         4   \n",
       "20       0.497585        False      adam         [5e-5,1e-4]  run_1         6   \n",
       "21       0.533816        False       sgd         [5e-5,1e-4]  run_1         4   \n",
       "22       0.096618        False      adam         [1e-4,5e-4]  run_1         4   \n",
       "23       0.659420        False       sgd         [1e-4,5e-4]  run_1         6   \n",
       "24       0.120773        False      adam         [5e-4,1e-3]  run_1         4   \n",
       "25       0.420290        False       sgd         [5e-4,1e-3]  run_1         4   \n",
       "26       0.183575        False      adam         [1e-3,5e-3]  run_1         5   \n",
       "27       0.096618        False       sgd         [1e-3,5e-3]  run_1         4   \n",
       "28       0.096618        False      adam         [5e-3,1e-2]  run_1         4   \n",
       "29       0.096618        False       sgd         [5e-3,1e-2]  run_1         4   \n",
       "..            ...          ...       ...                 ...    ...       ...   \n",
       "95       0.570000        False       sgd         [5e-5,1e-4]  run_1         4   \n",
       "96       0.342500        False      adam         [1e-4,5e-4]  run_1         6   \n",
       "97       0.607500        False       sgd         [1e-4,5e-4]  run_1         4   \n",
       "98       0.105000        False      adam         [5e-4,1e-3]  run_1         4   \n",
       "99       0.710000        False       sgd         [5e-4,1e-3]  run_1         7   \n",
       "100      0.105000        False      adam         [1e-3,5e-3]  run_1         4   \n",
       "101      0.640000        False       sgd         [1e-3,5e-3]  run_1         9   \n",
       "102      0.377500        False      adam         [5e-6,1e-5]  run_1         4   \n",
       "103      0.642032        False      adam                 NaN  run_2         4   \n",
       "104      0.193995        False       sgd                 NaN  run_2         4   \n",
       "105      0.764434        False      adam                 NaN  run_2         5   \n",
       "106      0.732102        False       sgd                 NaN  run_2         8   \n",
       "107      0.655889        False      adam                 NaN  run_2         4   \n",
       "108      0.475751        False       sgd                 NaN  run_2         4   \n",
       "109      0.644342        False      adam                 NaN  run_2        15   \n",
       "110      0.168591        False       sgd                 NaN  run_2         4   \n",
       "111      0.764434        False      adam                 NaN  run_2         9   \n",
       "112      0.554273        False       sgd                 NaN  run_2        10   \n",
       "113      0.568129        False      adam                 NaN  run_2         6   \n",
       "114      0.764434        False       sgd                 NaN  run_2        14   \n",
       "115      0.452703        False      adam                 NaN  run_2         4   \n",
       "116      0.234234        False       sgd                 NaN  run_2         4   \n",
       "117      0.750000        False      adam                 NaN  run_2         6   \n",
       "118      0.578829        False       sgd                 NaN  run_2         4   \n",
       "119      0.500000        False      adam                 NaN  run_2         4   \n",
       "120      0.524775        False       sgd                 NaN  run_2         4   \n",
       "121      0.707207        False      adam                 NaN  run_2        10   \n",
       "122      0.150901        False       sgd                 NaN  run_2         4   \n",
       "123      0.650901        False      adam                 NaN  run_2        10   \n",
       "124      0.529279        False       sgd                 NaN  run_2         4   \n",
       "\n",
       "     final_loss final_accuracy final_val_loss final_val_accuracy  \n",
       "0      0.502312       0.947025      11.928421           0.239024  \n",
       "1      0.311288       0.967400      11.619101           0.263415  \n",
       "2      4.411082       0.719641      12.300757           0.221951  \n",
       "3      9.715977       0.392013      16.118097           0.000000  \n",
       "4     14.633709       0.092095      16.118097           0.000000  \n",
       "5     14.357848       0.109209      16.118097           0.000000  \n",
       "6      4.070631       0.652827      16.028574           0.002646  \n",
       "7      1.477193       0.879859      16.117563           0.000000  \n",
       "8      5.677927       0.639576      16.118097           0.000000  \n",
       "9     10.009441       0.376325      16.118097           0.000000  \n",
       "10    11.638678       0.275618      16.118097           0.000000  \n",
       "11    14.409465       0.106007      16.118097           0.000000  \n",
       "12     2.949758       0.725424      16.100197           0.000000  \n",
       "13     2.432044       0.826271      16.118097           0.000000  \n",
       "14     1.248966       0.896610      16.118097           0.000000  \n",
       "15     8.976806       0.438983      16.118097           0.000000  \n",
       "16    13.536469       0.160169      16.118097           0.000000  \n",
       "17    14.560925       0.096610      16.118097           0.000000  \n",
       "18     2.081406       0.797502      15.511462           0.034759  \n",
       "19     3.698544       0.727922      16.118097           0.000000  \n",
       "20     5.101280       0.679348      13.880275           0.133550  \n",
       "21     4.101719       0.709783      16.118097           0.000000  \n",
       "22    14.068295       0.127174      16.118097           0.000000  \n",
       "23     1.035343       0.926087      13.074652           0.188925  \n",
       "24    13.542705       0.159783      16.118097           0.000000  \n",
       "25     5.632835       0.641304      16.118097           0.000000  \n",
       "26    13.490146       0.163043      13.387996           0.169381  \n",
       "27    14.068295       0.127174      16.118097           0.000000  \n",
       "28    14.033256       0.129348      16.118097           0.000000  \n",
       "29    14.068296       0.127174      16.118097           0.000000  \n",
       "..          ...            ...            ...                ...  \n",
       "95     3.775522       0.680899      16.103250           0.000000  \n",
       "96     8.938577       0.442697      14.652818           0.090909  \n",
       "97     2.770390       0.802247      13.964347           0.127946  \n",
       "98    13.854322       0.140449      16.118100           0.000000  \n",
       "99     0.777026       0.938202      14.086784           0.124579  \n",
       "100   13.836212       0.141573      16.118100           0.000000  \n",
       "101    2.322723       0.852809      13.784502           0.144781  \n",
       "102    7.517145       0.361798      14.940201           0.033670  \n",
       "103    2.946247       0.735810      16.089482           0.000000  \n",
       "104   10.925501       0.202270      15.011213           0.012384  \n",
       "105    0.295404       0.974200      12.694006           0.204334  \n",
       "106    0.475986       0.942208      13.152557           0.173375  \n",
       "107    1.855424       0.875129      16.118100           0.000000  \n",
       "108    6.235126       0.574819      16.118100           0.000000  \n",
       "109    2.279800       0.829721      12.834111           0.195046  \n",
       "110   11.222563       0.175439      15.410839           0.006192  \n",
       "111    0.230859       0.981424      12.996278           0.188854  \n",
       "112    3.820022       0.703818      13.087998           0.160991  \n",
       "113    5.352686       0.652219      13.468336           0.154799  \n",
       "114    0.125814       0.984520      13.115022           0.173375  \n",
       "115    6.168376       0.548028      16.026630           0.000000  \n",
       "116    9.926596       0.230536      15.771304           0.000000  \n",
       "117    0.403706       0.970677      12.921979           0.178788  \n",
       "118    3.807181       0.656218      16.117965           0.000000  \n",
       "119    5.519053       0.648129      16.118100           0.000000  \n",
       "120    4.053692       0.683519      16.118100           0.000000  \n",
       "121    0.552625       0.931244      13.424177           0.142424  \n",
       "122   11.656408       0.140546      15.454955           0.012121  \n",
       "123    2.354338       0.852376      13.006859           0.181818  \n",
       "124    4.938865       0.587462      16.111250           0.000000  \n",
       "\n",
       "[125 rows x 25 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation['final_loss'] = df_evaluation['epochs_losses'].apply(lambda x: x[-1])\n",
    "df_evaluation['final_accuracy'] = df_evaluation['epochs_accuracies'].apply(lambda x: x[-1])\n",
    "df_evaluation['final_val_loss'] = df_evaluation['epochs_val_losses'].apply(lambda x: x[-1])\n",
    "df_evaluation['final_val_accuracy'] = df_evaluation['epochs_val_accuracies'].apply(lambda x: x[-1])\n",
    "df_evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:5: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"\n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:7: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  import sys\n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:9: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  if __name__ == '__main__':\n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:10: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  # Remove the CWD from sys.path while we load stuff.\n",
      "c:\\python\\python36\\lib\\site-packages\\ipykernel_launcher.py:11: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "df_evaluation['learning_rate_categ'] = ''\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>5e-6)&(df_evaluation['learning_rate']<1e-5)] = '[5e-6,1e-5]'\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>1e-5)&(df_evaluation['learning_rate']<5e-5)] = '[1e-5,5e-5]'\n",
    "\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>5e-5)&(df_evaluation['learning_rate']<1e-4)] = '[5e-5,1e-4]'\n",
    "\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>1e-4)&(df_evaluation['learning_rate']<5e-4)] = '[1e-4,5e-4]'\n",
    "\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>5e-4)&(df_evaluation['learning_rate']<1e-3)] = '[5e-4,1e-3]'\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>1e-3)&(df_evaluation['learning_rate']<5e-3)] = '[1e-3,5e-3]'\n",
    "df_evaluation['learning_rate_categ'][(df_evaluation['learning_rate']>5e-3)&(df_evaluation['learning_rate']<1e-2)] = '[5e-3,1e-2]'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['[1e-3,5e-3]',\n",
       " '[1e-4,5e-4]',\n",
       " '[1e-5,5e-5]',\n",
       " '[5e-3,1e-2]',\n",
       " '[5e-4,1e-3]',\n",
       " '[5e-5,1e-4]',\n",
       " '[5e-6,1e-5]']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(np.unique(df_evaluation['learning_rate_categ']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">II) Vizualizing the accuracy depending on the parameters</font>**\n",
    "</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bb90b978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,10))\n",
    "plt.title('accuracy boxplots depending on the optimizers (x axis) and the learning rates intervals (colors)')\n",
    "sns.boxplot(x=\"optimizer\", y=\"test_accuracy\",\n",
    "            hue=\"learning_rate_categ\", palette=[\"red\",\"orange\", \"g\",\"b\",\"darkred\",\"violet\",\"black\"],\n",
    "            data=df_evaluation)\n",
    "sns.despine(offset=10, trim=True)\n",
    "#We see, thanks to this distributions, that the more interesting learning rates for our problem are in [5e-6,5e-4]\n",
    "#In effect, beyond this interval, final accuracies on the test are low\n",
    "\n",
    "#We can also observe that the sgd optimizer is better in  [1e-5, 5e-4] (highest learning rates of our interval), \n",
    "#whereas adam is better in [5e-6,5e-5] (lowest LR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bda86668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,10))\n",
    "plt.title('accuracy boxplots depending on the optimizers (x axis) and the batch sizes (colors)')\n",
    "sns.boxplot(x=\"optimizer\", y=\"test_accuracy\",\n",
    "            hue=\"batch_size\", palette=[\"red\",\"orange\", \"g\"],\n",
    "            data=df_evaluation)\n",
    "sns.despine(offset=10, trim=True)\n",
    "#With a batch size of \"100\", we can observe very huge gaps between accuracy extrema\n",
    "#Even if there is a huge gap with a batch of 300 on the sgd optimizer, this batch size is interesting because it seems to \n",
    "##generate the best of all the accuracies on the amam optimizer\n",
    "\n",
    "#The batch size of 200 seems to be correct for both sgd and adam optimizers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bb956940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,10))\n",
    "plt.title('accuracy boxplots depending on the number races (x axis) and the optimizers (colors)')\n",
    "sns.boxplot(x=\"n_races\", y=\"test_accuracy\",\n",
    "            hue=\"optimizer\", palette=[\"red\",\"orange\"],\n",
    "            data=df_evaluation)\n",
    "sns.despine(offset=10, trim=True)\n",
    "#with all batch sizes included in the analyzis of the accuracy distribution, we can see that sgd is more stable than adam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bddaa7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,10))\n",
    "plt.title('Number of epochs depending on the optimizers (x axis) and the learning rates intervals (colors)')\n",
    "sns.boxplot(x=\"optimizer\", y=\"n_epochs\",\n",
    "            hue=\"learning_rate_categ\", palette=[\"red\",\"orange\", \"g\",\"b\",\"darkred\",\"violet\",\"black\"],\n",
    "            data=df_evaluation)\n",
    "sns.despine(offset=10, trim=True)\n",
    "#This plot confirms what we observed before : the best learning rates are in [5e-6,5e-4] ; out of this interval epochs tends to\n",
    "## be close to 4 (which is the minimum number of epochs if the validation loss does'nt move)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1b5bdd987b8>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b59b2d4f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_evaluation.plot.scatter(x='final_loss',\n",
    "                           y='final_val_loss',\n",
    "                           c='DarkBlue',\n",
    "                           title = 'Loss on validation data depending on the accuracy on the train data, at the last epoch',\n",
    "                           figsize=(15,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1b5bdd0c390>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bdcdbc50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_evaluation.plot.scatter(x='final_accuracy',\n",
    "                           y='final_val_accuracy',\n",
    "                           c='DarkRed',\n",
    "                           title = 'Accuracy on validation data depending on the accuracy on the train data, at the last epoch',\n",
    "                           figsize=(15,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>races</th>\n",
       "      <th>training_len</th>\n",
       "      <th>testing_len</th>\n",
       "      <th>iter_number</th>\n",
       "      <th>n_races</th>\n",
       "      <th>random_selection_number</th>\n",
       "      <th>batch_size</th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>fitting_time</th>\n",
       "      <th>prediction_time</th>\n",
       "      <th>...</th>\n",
       "      <th>test_accuracy</th>\n",
       "      <th>is_augmented</th>\n",
       "      <th>optimizer</th>\n",
       "      <th>learning_rate_categ</th>\n",
       "      <th>run</th>\n",
       "      <th>n_epochs</th>\n",
       "      <th>final_loss</th>\n",
       "      <th>final_accuracy</th>\n",
       "      <th>final_val_loss</th>\n",
       "      <th>final_val_accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>3077</td>\n",
       "      <td>125</td>\n",
       "      <td>...</td>\n",
       "      <td>0.722826</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>8</td>\n",
       "      <td>0.502312</td>\n",
       "      <td>0.947025</td>\n",
       "      <td>11.928421</td>\n",
       "      <td>0.239024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000089</td>\n",
       "      <td>3454</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.733696</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>9</td>\n",
       "      <td>0.311288</td>\n",
       "      <td>0.967400</td>\n",
       "      <td>11.619101</td>\n",
       "      <td>0.263415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>['afghan_hound', 'boxer', 'cairn', 'collie', '...</td>\n",
       "      <td>1637</td>\n",
       "      <td>552</td>\n",
       "      <td>2</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000443</td>\n",
       "      <td>2209</td>\n",
       "      <td>122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.543478</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>4.411082</td>\n",
       "      <td>0.719641</td>\n",
       "      <td>12.300757</td>\n",
       "      <td>0.221951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1672</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.497585</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>5.101280</td>\n",
       "      <td>0.679348</td>\n",
       "      <td>13.880275</td>\n",
       "      <td>0.133550</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000428</td>\n",
       "      <td>1639</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.659420</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>1.035343</td>\n",
       "      <td>0.926087</td>\n",
       "      <td>13.074652</td>\n",
       "      <td>0.188925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>['black_and_tan_coonhound', 'briard', 'bull_ma...</td>\n",
       "      <td>1227</td>\n",
       "      <td>414</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>1402</td>\n",
       "      <td>90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.183575</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>5</td>\n",
       "      <td>13.490146</td>\n",
       "      <td>0.163043</td>\n",
       "      <td>13.387996</td>\n",
       "      <td>0.169381</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>['american_staffordshire_terrier', 'appenzelle...</td>\n",
       "      <td>1298</td>\n",
       "      <td>437</td>\n",
       "      <td>21</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000086</td>\n",
       "      <td>1726</td>\n",
       "      <td>95</td>\n",
       "      <td>...</td>\n",
       "      <td>0.713959</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>0.323724</td>\n",
       "      <td>0.958890</td>\n",
       "      <td>13.547871</td>\n",
       "      <td>0.153846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>['airedale', 'black_and_tan_coonhound', 'bluet...</td>\n",
       "      <td>1208</td>\n",
       "      <td>407</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000059</td>\n",
       "      <td>1650</td>\n",
       "      <td>89</td>\n",
       "      <td>...</td>\n",
       "      <td>0.636364</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>2.801164</td>\n",
       "      <td>0.818985</td>\n",
       "      <td>12.663454</td>\n",
       "      <td>0.208609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>['airedale', 'black_and_tan_coonhound', 'bluet...</td>\n",
       "      <td>1208</td>\n",
       "      <td>407</td>\n",
       "      <td>31</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000059</td>\n",
       "      <td>2944</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.742015</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>11</td>\n",
       "      <td>0.117741</td>\n",
       "      <td>0.986755</td>\n",
       "      <td>12.543878</td>\n",
       "      <td>0.221854</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>['border_terrier', 'bouvier_des_flandres', 'da...</td>\n",
       "      <td>1285</td>\n",
       "      <td>433</td>\n",
       "      <td>14</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000026</td>\n",
       "      <td>1746</td>\n",
       "      <td>94</td>\n",
       "      <td>...</td>\n",
       "      <td>0.734411</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>6</td>\n",
       "      <td>0.230336</td>\n",
       "      <td>0.977155</td>\n",
       "      <td>12.233704</td>\n",
       "      <td>0.236025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>['border_terrier', 'bouvier_des_flandres', 'da...</td>\n",
       "      <td>1285</td>\n",
       "      <td>433</td>\n",
       "      <td>15</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000026</td>\n",
       "      <td>2278</td>\n",
       "      <td>94</td>\n",
       "      <td>...</td>\n",
       "      <td>0.593533</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>8</td>\n",
       "      <td>3.858888</td>\n",
       "      <td>0.696781</td>\n",
       "      <td>12.581742</td>\n",
       "      <td>0.177019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>['border_terrier', 'bouvier_des_flandres', 'da...</td>\n",
       "      <td>1285</td>\n",
       "      <td>433</td>\n",
       "      <td>17</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000056</td>\n",
       "      <td>3445</td>\n",
       "      <td>95</td>\n",
       "      <td>...</td>\n",
       "      <td>0.734411</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>12</td>\n",
       "      <td>0.482152</td>\n",
       "      <td>0.952233</td>\n",
       "      <td>12.140666</td>\n",
       "      <td>0.239130</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>['border_terrier', 'bouvier_des_flandres', 'da...</td>\n",
       "      <td>1285</td>\n",
       "      <td>433</td>\n",
       "      <td>19</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0.000438</td>\n",
       "      <td>1991</td>\n",
       "      <td>94</td>\n",
       "      <td>...</td>\n",
       "      <td>0.736721</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>7</td>\n",
       "      <td>0.865658</td>\n",
       "      <td>0.941848</td>\n",
       "      <td>11.963432</td>\n",
       "      <td>0.257764</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>6819</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.707500</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-6,1e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>20</td>\n",
       "      <td>0.189719</td>\n",
       "      <td>0.974157</td>\n",
       "      <td>13.801976</td>\n",
       "      <td>0.131313</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000029</td>\n",
       "      <td>1832</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.685000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>5</td>\n",
       "      <td>0.485087</td>\n",
       "      <td>0.951685</td>\n",
       "      <td>14.200537</td>\n",
       "      <td>0.111111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000055</td>\n",
       "      <td>3204</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>9</td>\n",
       "      <td>2.231899</td>\n",
       "      <td>0.858427</td>\n",
       "      <td>13.839666</td>\n",
       "      <td>0.141414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000291</td>\n",
       "      <td>1580</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.607500</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-4,5e-4]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>4</td>\n",
       "      <td>2.770390</td>\n",
       "      <td>0.802247</td>\n",
       "      <td>13.964347</td>\n",
       "      <td>0.127946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000604</td>\n",
       "      <td>2604</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.710000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-4,1e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>7</td>\n",
       "      <td>0.777026</td>\n",
       "      <td>0.938202</td>\n",
       "      <td>14.086784</td>\n",
       "      <td>0.124579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'che...</td>\n",
       "      <td>1187</td>\n",
       "      <td>400</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>3042</td>\n",
       "      <td>88</td>\n",
       "      <td>...</td>\n",
       "      <td>0.640000</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-3,5e-3]</td>\n",
       "      <td>run_1</td>\n",
       "      <td>9</td>\n",
       "      <td>2.322723</td>\n",
       "      <td>0.852809</td>\n",
       "      <td>13.784502</td>\n",
       "      <td>0.144781</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>2015</td>\n",
       "      <td>95</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>5</td>\n",
       "      <td>0.295404</td>\n",
       "      <td>0.974200</td>\n",
       "      <td>12.694006</td>\n",
       "      <td>0.204334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000049</td>\n",
       "      <td>3112</td>\n",
       "      <td>97</td>\n",
       "      <td>...</td>\n",
       "      <td>0.732102</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>8</td>\n",
       "      <td>0.475986</td>\n",
       "      <td>0.942208</td>\n",
       "      <td>13.152557</td>\n",
       "      <td>0.173375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>8158</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.644342</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-6,1e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>15</td>\n",
       "      <td>2.279800</td>\n",
       "      <td>0.829721</td>\n",
       "      <td>12.834111</td>\n",
       "      <td>0.195046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>8</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>4922</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>9</td>\n",
       "      <td>0.230859</td>\n",
       "      <td>0.981424</td>\n",
       "      <td>12.996278</td>\n",
       "      <td>0.188854</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>5297</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.554273</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>3.820022</td>\n",
       "      <td>0.703818</td>\n",
       "      <td>13.087998</td>\n",
       "      <td>0.160991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>3212</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.568129</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>6</td>\n",
       "      <td>5.352686</td>\n",
       "      <td>0.652219</td>\n",
       "      <td>13.468336</td>\n",
       "      <td>0.154799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>['brabancon_griffon', 'brittany_spaniel', 'bul...</td>\n",
       "      <td>1292</td>\n",
       "      <td>433</td>\n",
       "      <td>11</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>7417</td>\n",
       "      <td>96</td>\n",
       "      <td>...</td>\n",
       "      <td>0.764434</td>\n",
       "      <td>False</td>\n",
       "      <td>sgd</td>\n",
       "      <td>[5e-5,1e-4]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>14</td>\n",
       "      <td>0.125814</td>\n",
       "      <td>0.984520</td>\n",
       "      <td>13.115022</td>\n",
       "      <td>0.173375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>14</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>200</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>2393</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>6</td>\n",
       "      <td>0.403706</td>\n",
       "      <td>0.970677</td>\n",
       "      <td>12.921979</td>\n",
       "      <td>0.178788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>18</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>4357</td>\n",
       "      <td>98</td>\n",
       "      <td>...</td>\n",
       "      <td>0.707207</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[5e-6,1e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>0.552625</td>\n",
       "      <td>0.931244</td>\n",
       "      <td>13.424177</td>\n",
       "      <td>0.142424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>['airedale', 'american_staffordshire_terrier',...</td>\n",
       "      <td>1319</td>\n",
       "      <td>444</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>4380</td>\n",
       "      <td>99</td>\n",
       "      <td>...</td>\n",
       "      <td>0.650901</td>\n",
       "      <td>False</td>\n",
       "      <td>adam</td>\n",
       "      <td>[1e-5,5e-5]</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "      <td>2.354338</td>\n",
       "      <td>0.852376</td>\n",
       "      <td>13.006859</td>\n",
       "      <td>0.181818</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>29 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 races  training_len  \\\n",
       "0    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "1    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "2    ['afghan_hound', 'boxer', 'cairn', 'collie', '...          1637   \n",
       "20   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "23   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "26   ['black_and_tan_coonhound', 'briard', 'bull_ma...          1227   \n",
       "41   ['american_staffordshire_terrier', 'appenzelle...          1298   \n",
       "50   ['airedale', 'black_and_tan_coonhound', 'bluet...          1208   \n",
       "51   ['airedale', 'black_and_tan_coonhound', 'bluet...          1208   \n",
       "74   ['border_terrier', 'bouvier_des_flandres', 'da...          1285   \n",
       "75   ['border_terrier', 'bouvier_des_flandres', 'da...          1285   \n",
       "77   ['border_terrier', 'bouvier_des_flandres', 'da...          1285   \n",
       "79   ['border_terrier', 'bouvier_des_flandres', 'da...          1285   \n",
       "90   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "92   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "94   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "97   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "99   ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "101  ['brabancon_griffon', 'brittany_spaniel', 'che...          1187   \n",
       "105  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "106  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "109  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "111  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "112  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "113  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "114  ['brabancon_griffon', 'brittany_spaniel', 'bul...          1292   \n",
       "117  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "121  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "123  ['airedale', 'american_staffordshire_terrier',...          1319   \n",
       "\n",
       "     testing_len  iter_number  n_races  random_selection_number  batch_size  \\\n",
       "0            552            0       12                        0         100   \n",
       "1            552            1       12                        0         100   \n",
       "2            552            2       12                        0         100   \n",
       "20           414            0       10                        0         100   \n",
       "23           414            3       10                        0         100   \n",
       "26           414            6       10                        0         100   \n",
       "41           437           21       10                        2         100   \n",
       "50           407           30       10                        3         100   \n",
       "51           407           31       10                        3         100   \n",
       "74           433           14       10                        1         100   \n",
       "75           433           15       10                        1         100   \n",
       "77           433           17       10                        1         100   \n",
       "79           433           19       10                        1         100   \n",
       "90           400            0       10                        0         200   \n",
       "92           400            2       10                        0         200   \n",
       "94           400            4       10                        0         200   \n",
       "97           400            7       10                        0         200   \n",
       "99           400            9       10                        0         200   \n",
       "101          400           11       10                        0         200   \n",
       "105          433            2       10                        0         200   \n",
       "106          433            3       10                        0         200   \n",
       "109          433            6       10                        0         300   \n",
       "111          433            8       10                        0         300   \n",
       "112          433            9       10                        0         300   \n",
       "113          433           10       10                        0         300   \n",
       "114          433           11       10                        0         300   \n",
       "117          444           14       10                        1         200   \n",
       "121          444           18       10                        1         300   \n",
       "123          444           20       10                        1         300   \n",
       "\n",
       "     learning_rate  fitting_time  prediction_time        ...          \\\n",
       "0         0.000046          3077              125        ...           \n",
       "1         0.000089          3454              122        ...           \n",
       "2         0.000443          2209              122        ...           \n",
       "20        0.000065          1672               90        ...           \n",
       "23        0.000428          1639               90        ...           \n",
       "26        0.004850          1402               90        ...           \n",
       "41        0.000086          1726               95        ...           \n",
       "50        0.000059          1650               89        ...           \n",
       "51        0.000059          2944               88        ...           \n",
       "74        0.000026          1746               94        ...           \n",
       "75        0.000026          2278               94        ...           \n",
       "77        0.000056          3445               95        ...           \n",
       "79        0.000438          1991               94        ...           \n",
       "90        0.000005          6819               88        ...           \n",
       "92        0.000029          1832               88        ...           \n",
       "94        0.000055          3204               88        ...           \n",
       "97        0.000291          1580               88        ...           \n",
       "99        0.000604          2604               88        ...           \n",
       "101       0.001061          3042               88        ...           \n",
       "105       0.000049          2015               95        ...           \n",
       "106       0.000049          3112               97        ...           \n",
       "109       0.000006          8158               96        ...           \n",
       "111       0.000042          4922               96        ...           \n",
       "112       0.000042          5297               96        ...           \n",
       "113       0.000091          3212               96        ...           \n",
       "114       0.000091          7417               96        ...           \n",
       "117       0.000047          2393               99        ...           \n",
       "121       0.000009          4357               98        ...           \n",
       "123       0.000050          4380               99        ...           \n",
       "\n",
       "    test_accuracy is_augmented optimizer learning_rate_categ    run  n_epochs  \\\n",
       "0        0.722826        False       sgd         [1e-5,5e-5]  run_1         8   \n",
       "1        0.733696        False       sgd         [5e-5,1e-4]  run_1         9   \n",
       "2        0.543478        False       sgd         [1e-4,5e-4]  run_1         6   \n",
       "20       0.497585        False      adam         [5e-5,1e-4]  run_1         6   \n",
       "23       0.659420        False       sgd         [1e-4,5e-4]  run_1         6   \n",
       "26       0.183575        False      adam         [1e-3,5e-3]  run_1         5   \n",
       "41       0.713959        False       sgd         [5e-5,1e-4]  run_1         6   \n",
       "50       0.636364        False      adam         [5e-5,1e-4]  run_1         6   \n",
       "51       0.742015        False       sgd         [5e-5,1e-4]  run_1        11   \n",
       "74       0.734411        False      adam         [1e-5,5e-5]  run_1         6   \n",
       "75       0.593533        False       sgd         [1e-5,5e-5]  run_1         8   \n",
       "77       0.734411        False       sgd         [5e-5,1e-4]  run_1        12   \n",
       "79       0.736721        False       sgd         [1e-4,5e-4]  run_1         7   \n",
       "90       0.707500        False      adam         [5e-6,1e-5]  run_1        20   \n",
       "92       0.685000        False      adam         [1e-5,5e-5]  run_1         5   \n",
       "94       0.660000        False      adam         [5e-5,1e-4]  run_1         9   \n",
       "97       0.607500        False       sgd         [1e-4,5e-4]  run_1         4   \n",
       "99       0.710000        False       sgd         [5e-4,1e-3]  run_1         7   \n",
       "101      0.640000        False       sgd         [1e-3,5e-3]  run_1         9   \n",
       "105      0.764434        False      adam         [1e-5,5e-5]  run_2         5   \n",
       "106      0.732102        False       sgd         [1e-5,5e-5]  run_2         8   \n",
       "109      0.644342        False      adam         [5e-6,1e-5]  run_2        15   \n",
       "111      0.764434        False      adam         [1e-5,5e-5]  run_2         9   \n",
       "112      0.554273        False       sgd         [1e-5,5e-5]  run_2        10   \n",
       "113      0.568129        False      adam         [5e-5,1e-4]  run_2         6   \n",
       "114      0.764434        False       sgd         [5e-5,1e-4]  run_2        14   \n",
       "117      0.750000        False      adam         [1e-5,5e-5]  run_2         6   \n",
       "121      0.707207        False      adam         [5e-6,1e-5]  run_2        10   \n",
       "123      0.650901        False      adam         [1e-5,5e-5]  run_2        10   \n",
       "\n",
       "     final_loss final_accuracy final_val_loss final_val_accuracy  \n",
       "0      0.502312       0.947025      11.928421           0.239024  \n",
       "1      0.311288       0.967400      11.619101           0.263415  \n",
       "2      4.411082       0.719641      12.300757           0.221951  \n",
       "20     5.101280       0.679348      13.880275           0.133550  \n",
       "23     1.035343       0.926087      13.074652           0.188925  \n",
       "26    13.490146       0.163043      13.387996           0.169381  \n",
       "41     0.323724       0.958890      13.547871           0.153846  \n",
       "50     2.801164       0.818985      12.663454           0.208609  \n",
       "51     0.117741       0.986755      12.543878           0.221854  \n",
       "74     0.230336       0.977155      12.233704           0.236025  \n",
       "75     3.858888       0.696781      12.581742           0.177019  \n",
       "77     0.482152       0.952233      12.140666           0.239130  \n",
       "79     0.865658       0.941848      11.963432           0.257764  \n",
       "90     0.189719       0.974157      13.801976           0.131313  \n",
       "92     0.485087       0.951685      14.200537           0.111111  \n",
       "94     2.231899       0.858427      13.839666           0.141414  \n",
       "97     2.770390       0.802247      13.964347           0.127946  \n",
       "99     0.777026       0.938202      14.086784           0.124579  \n",
       "101    2.322723       0.852809      13.784502           0.144781  \n",
       "105    0.295404       0.974200      12.694006           0.204334  \n",
       "106    0.475986       0.942208      13.152557           0.173375  \n",
       "109    2.279800       0.829721      12.834111           0.195046  \n",
       "111    0.230859       0.981424      12.996278           0.188854  \n",
       "112    3.820022       0.703818      13.087998           0.160991  \n",
       "113    5.352686       0.652219      13.468336           0.154799  \n",
       "114    0.125814       0.984520      13.115022           0.173375  \n",
       "117    0.403706       0.970677      12.921979           0.178788  \n",
       "121    0.552625       0.931244      13.424177           0.142424  \n",
       "123    2.354338       0.852376      13.006859           0.181818  \n",
       "\n",
       "[29 rows x 25 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "focus = df_evaluation[df_evaluation['final_val_accuracy']>=.1]\n",
    "focus"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p>\n",
    "        **<font color='#D2691E'size=\"4\">II) Vizualizing the losses and accuracies depending on the training epochs</font>**\n",
    "</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[11.177390171632402,\n",
       " 5.857689602159346,\n",
       " 2.6551777008790354,\n",
       " 1.5934128820920437,\n",
       " 1.1124129741290962,\n",
       " 0.7947726992202563,\n",
       " 0.7803367870361123,\n",
       " 0.5023115000900019]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation['epochs_losses'][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[15.50204170040968,\n",
       " 15.00998678440001,\n",
       " 12.279677228229803,\n",
       " 12.089079135801734,\n",
       " 11.908222157780717,\n",
       " 12.508801669609255,\n",
       " 12.06607127771145,\n",
       " 11.928420572746091]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_evaluation['epochs_val_losses'][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1b5bdacbef0>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bdc19ba8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,8))\n",
    "plt.title(\"training and validation losses, depending on the epochs :\")\n",
    "x1 = df_evaluation['epochs_losses'][0]\n",
    "x2 = df_evaluation['epochs_val_losses'][0]\n",
    "y = range(0,len(x1))\n",
    "\n",
    "plt.plot(y,x1,label='training loss')\n",
    "plt.plot(y,x2,label='validation loss')\n",
    "plt.legend(loc='upper right')\n",
    "#The training loss is less important than the validation one, which is good : what about the accuracy ? "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1b5bdb54080>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAHiCAYAAAB7iyTuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzs3Xd8leX9//H3lUU22RBWwhQIm4CAC4rgaMUWB9aBgBNX+3W0jtZZW+uXWmur7Q8tCAgIzrpQ61epWkHCEERAEQgkhJEQQkIGGef6/XHfiSchIQESTsbr2Ucenpx7fe77nEPPO9d1X5ex1goAAAAA0Dz5+boAAAAAAEDdCG0AAAAA0IwR2gAAAACgGSO0AQAAAEAzRmgDAAAAgGaM0AYAAAAAzRihDUCTMcb8wxjz28Ze15eMMcuNMdc3wX7TjTHnuo/vN8a80JB1T+A4Zxljvj3ROlsrY8xhY0yPU3xMa4zpdSqPWR9jzDRjzOdev5/y69IYmuO1baiarwEASFKArwsA0DwZY9IlXW+t/ehE92Gtvbkp1m3trLW/b6x9GWOspN7W2u/dfX8m6bTG2n9rYa0N93UNzVFLuC7GmOWSXrLW1vmHDgBo6WhpA3BCjDH80QfNBu9HAEBrRmgDcBRjzAJJ3SS97XaP+pUxJtntcnSdMWaXpI/ddV8xxuw1xhwyxnxqjEnx2s+LxpjfuY/HGmMyjTF3GWP2G2P2GGOmn+C6scaYt40x+caYNGPM747VnagBNT5rjHnXGFNgjPnSGNPTa/kEY8wWd9u/STJ1HKOTMabYGBPj9dxQY0yOMSbQGNPTGPOxMeaA+9xCY0xUHft62Bjzktfv1xhjdrrbPlBj3ZHGmBXGmDz3Ov3NGBPkLvvUXW29+zpOqby2Xtv3c7t85hljvjHGTGrotTnO6xxijPmTex6HjDGfG2NC3GVnGmO+cGvIMMZMc5+v1hW1lq571hhzqzFmq6St7nN/cfeRb4xZY4w5y2t9f+N0Pd3mns8aY0xXr331ch+3M8bMMsbsMsbsM07X3cpa44wx77i15hpjPjPGNOj/S40x97ivUZYxZkaNZcc6ZuXn4X73vZNujLnqOLc91mfpLfd6rZLUs0Zd3telvs/KRGPMt+7r+5wx5j+mjq7Ebs1Pu9ciy33criE119jP45LOkvQ39z3+N6/F5xpjthpjDrp1G6/tZhhjNrvLPjDGJB3jdRvl9f5cb4wZ67VsuTHmD8aYVe55/8tU/zdgknE+V3nuuv28lnU1xrxujMk2zmf7bzWOO8utb4cx5gKv56cZY7a7r8EO7/cCgNaL0AbgKNbaayTtknSRtTbcWvuk1+JzJPWTdJ77+zJJvSUlSForaeExdt1RUntJnSVdJ+lZY0z0Caz7rKRCd51r3Z9jqa/Gn0t6RFK0pO8lPS45X9AlvSbpN5LiJG2TdEZtB7DWZklaIekSr6evlPSqtbZMTtj7g6ROcq5fV0kP11O3jDH9Jf1d0jXutrGSunitUiHpf9z6RksaL+kWt6az3XUGu6/jkhr7DpT0tqQP5Vyb2yUtNMZ4d5+s9drU4VjXeZak4ZLGSIqR9CtJHmNMN3e7v0qKlzRE0lfHuiY1/FTS6ZL6u7+nufuIkbRI0ivGmGB32Z3u+VwoKVLSDElFtezzj5L6uPvpJec9+KC77C5JmW6tHSTdL8nWV6Qx5nxJd0uaIOca1bwn8VjHlJz3epz7/LWSZnu9Tg3Z9lifpRJJiXKuR7UwWYtjfVZelXSfnPfot3Je67o8IGmUW/NgSSPlfM4aUnMVa+0Dkj6TdJv7Hr/Na/FPJI1w93+53H+zjDE/lfO6TZbzOn4maXFtRRpjOkt6V9Lv5Lyn7pb0mjEm3mu1qXKuWydJ5ZKecbft4+73l+5x3pPzh7AgY4y/pHck7ZSU7J7ny177PF3ONYyT9KSkfxpHmLv/C6y1EXKuca2fF+P8MSSvtmUAWiBrLT/88MPPUT+S0iWd6/V7spwvpz2OsU2Uu0579/cXJf3OfTxWUrGkAK/190sadTzrSvKXVCbpNK9lv5P0eQPPq7YaX/BafqGkLe7jqZJWei0zcr6wX1/Hvq+X9LHXuhmSzq5j3Z9KWlfb9ZYT5l5yHz8o6WWv9cIklXq/NjX2+0tJb3j9biX18vp9rKRM9/FZkvZK8vNavljSw/Vdm+O5znL+QFgsJzzWXO8+73prLFvufa0lTfN+nd39/6ieOg5WHlfOl+CL61jPygk8Rs4fBHp6LRstaYf7+FFJ//K+pg28HnMkPeH1e5/jOOZYOWEgzGv5Ukm/beC29X2W+not+30t17hXAz8rK2p8VjJU92dlm6QLvX4/T1J6fTU35H3iVfeZNa7Xve7jZZKu81rmJye8J9Wy719LWlDjuQ8kXet1bO/Xtb+cz6e/+/osrXGc3e75jZaU7X2ONd7n33v9HuqeT0c5n/88OX8cCjme9yA//PDTsn9oaQNwvDIqHxinu9kTxululi8neEjOX4drc8BaW+71e5GkugY6qGvdeDmDKGV4LfN+XE0Da9xbR02dvPdtrbXHOpaclobRxphOks6W80XrM7eOBGPMy8aY3W4dL6nu6+StZg2Fkg54nV8f43TX2+vu9/cN3G/Vvq21Hq/ndsr5q3+luq5NNfVc5zhJwXK+qNfUtY7nG6ra6+F2qdvsdlXLkxMaK69HQ44VL+dL8hq3S1uepPfd5yXpf+W0MH3odlG7t4F1Vnsd5Vznhh5Tkg66r7339p0auO3xfJa866rN8XxWMlW3TjWOVXk+9dV8POqqNUnSX7yuV66ckNlZR0uSdFnluu76Z8ppmaxU8/oFynnPVTtH93OW4R6nq6SdNc6x1tqttZWtweHue2CKpJsl7TFOV9W+dV0AAK0HoQ1AXerq8uX9/JWSLpbT1au9nNY4qY77vhpJtpxWB+8ugl2Psf7J1LjHe9/uPTF1Hstamyenq+Hl7nEXu19eJadrpJU0yFobKenqE6whVE73s0p/l7RFzgiRkXK6fTX0+mdJ6mqq35PVTU5rwPE61nXOkdMFr7b74TLqeF5yWpBCvX7vWMs6Ve9H49y/9ms51z/aWhsl6ZB+uB7HOlalHDmtPCnW2ij3p711R1G01hZYa++y1vaQdJGkO40x4+vZp1TjdZRznRt0TFe02zXOe/usBm5bl8rPUl11HY898vpMup+VLnWvriw5gcj7uFkneOx6u6fWkCHpJq/rFWWtDbHWflHHugtqrBtmrX3Ca52a169MzutS7Ry9/v3Y7e63mzmBAXSstR9YayfICY5bJD1/vPsA0PIQ2gDUZZ+k+uZnipB0RE7LT6icVp4mZa2tkPS6pIeNMaHuX5mnNlGN70pKMcZMdr9c3aHag4O3RW49l7iPves4LCnPvU/mngbW8Kqkn7j3pwTJ6Z7n/W93hKR8SYfdazGzxvbHeh2/lBOMfmWcwVLGygkiL9ex/rHUeZ3dFoY5kp4yzoAt/saY0cYZeGKhnAEjLjfGBBhnYIwh7qZfSZrsvs695NzbVF8N5XK7nRljHpRz71qlFyQ9Zozp7d4fNMgY4x2AK2t9XtKfjTEJknNfkzGm8n6onxhjerlfwPPl3FNY4S570RjzYh21LZU0zRjT3w3eDzX0mF4ece+HOkvO/VqvHMe2R6nls9Rf9d8fWpd3JQ00xvzU/azcqmN/VhZL+o0xJt69H+5BOa3PJ6Ih/1Z5+4ek+4w7UI4xpr0x5rI61n1J0kXGmPPc922wcQZK8Q6kV3u9ro/KuY+1Qs5r/mNjzHjj3D96l5zPyBeSVskJuk8YY8Lc/dZ6v6w3Y0wH4wxuEubu67Dc9x+A1o3QBqAuf5DzpSrPGHN3HevMl9P9Z7ekTZJWnqLabpPTmrNX0gI5XwCP1LHuCddorc2RdJmkJ+SEkd6S/lvPZm+56+2z1q73ev4RScPktPy8K+fLckNq+EbOF+BFcr7kHVT1bmd3y2nlKpDz5X1JjV08LGme+zpeXmPfpZImSbpATsvAc5KmWmu3NKS2Guq7zndL+lrOQCG5cgbP8LPW7pJzb9Rd7vNfyRk4QpL+LOf+oH2S5unYg9xIzr1GyyR959ZSoupd156S80X6QzmB65+SQmrZz6/ldIFc6Xb1/Eg/zG3X2/39sJyBZ56z1i53l3VVHe8Pa+0ySU/LGXX1e/e/DT2m5LzXD8ppvVko6Wav16m+bY/lNjndBvfKuWdtbgO3q8brs/KknM9Kf0mrVffn8nfu8g1y3hdr3edOxF8kXWqckRafaUCtb8h5/73sXq+Ncj4Dta2bIacF+X45fwzIkPMHF+/vTwvkXLu9croB3+Fu+62cFvW/yvl8XSRncKdSN9RdJOeexl1yPtNTGnCufnI+K1lyPi/nyB14qCZjzFnGmMMN2CeAFsD80HMHAFomY8wfJXW01p5oKwFwUtxW0PVyur+WNfK+x8oZmOZY3Q2bFbfLbaakq6y1n/i6nqZimNgbwClCSxuAFscY09ft2maMMSPldJt7w9d1oe1yW0/6NXZga0ncLoRRbrfXynsrT1XrOwC0asd9AywANAMRcrpEdpIzFPif5AzDDsB3Rsvpxhskp4vsT621xb4tCQBaB7pHAgAAAEAzRvdIAAAAAGjGCG0AAAAA0Iz57J62uLg4m5yc7KvDAwAAAIBPrVmzJsdaG1/fej4LbcnJyVq9erWvDg8AAAAAPmWM2dmQ9egeCQAAAADNGKENAAAAAJoxQhsAAAAANGNMrg0AAAA0orKyMmVmZqqkpMTXpaCZCA4OVpcuXRQYGHhC2xPaAAAAgEaUmZmpiIgIJScnyxjj63LgY9ZaHThwQJmZmerevfsJ7YPukQAAAEAjKikpUWxsLIENkiRjjGJjY0+q5ZXQBgAAADQyAhu8nez7gdAGAAAAtCJ5eXl67rnnTmjbCy+8UHl5ecdc58EHH9RHH310QvvHiSG0AQAAAK3IsUJbRUXFMbd97733FBUVdcx1Hn30UZ177rknXJ8vlJeX+7qEk0JoAwAAAFqRe++9V9u2bdOQIUN0zz33aPny5Ro3bpyuvPJKDRw4UJL005/+VMOHD1dKSopmz55dtW1ycrJycnKUnp6ufv366YYbblBKSoomTpyo4uJiSdK0adP06quvVq3/0EMPadiwYRo4cKC2bNkiScrOztaECRM0bNgw3XTTTUpKSlJOTs5Rtc6cOVOpqalKSUnRQw89VPV8WlqaxowZo8GDB2vkyJEqKChQRUWF7r77bg0cOFCDBg3SX//612o1S9Lq1as1duxYSdLDDz+sG2+8URMnTtTUqVOVnp6us846S8OGDdOwYcP0xRdfVB3vySef1MCBAzV48OCq6zds2LCq5Vu3btXw4cNP+rU5UYweCQAAADSRR97+Rpuy8ht1n/07Reqhi1LqXP7EE09o48aN+uqrryRJy5cv16pVq7Rx48aq0QvnzJmjmJgYFRcXa8SIEbrkkksUGxtbbT9bt27V4sWL9fzzz+vyyy/Xa6+9pquvvvqo48XFxWnt2rV67rnnNGvWLL3wwgt65JFH9KMf/Uj33Xef3n///WrB0Nvjjz+umJgYVVRUaPz48dqwYYP69u2rKVOmaMmSJRoxYoTy8/MVEhKi2bNna8eOHVq3bp0CAgKUm5tb77Vas2aNPv/8c4WEhKioqEj//ve/FRwcrK1bt+rnP/+5Vq9erWXLlunNN9/Ul19+qdDQUOXm5iomJkbt27fXV199pSFDhmju3LmaNm1avcdrKoQ2AAAAoJUbOXJkteHmn3nmGb3xxhuSpIyMDG3duvWo0Na9e3cNGTJEkjR8+HClp6fXuu/JkydXrfP6669Lkj7//POq/Z9//vmKjo6uddulS5dq9uzZKi8v1549e7Rp0yYZY5SYmKgRI0ZIkiIjIyVJH330kW6++WYFBDgRJiYmpt7znjRpkkJCQiQ58+fddttt+uqrr+Tv76/vvvuuar/Tp09XaGhotf1ef/31mjt3rp566iktWbJEq1atqvd4TYXQBgAAADSRY7WInUphYWFVj5cvX66PPvpIK1asUGhoqMaOHVvrcPTt2rWreuzv71/VPbKu9fz9/avuHbPW1lvTjh07NGvWLKWlpSk6OlrTpk1TSUmJrLW1jrZY1/MBAQHyeDySdNR5eJ/3n//8Z3Xo0EHr16+Xx+NRcHDwMfd7ySWXVLUYDh8+/KhQeypxTxsAAADQikRERKigoKDO5YcOHVJ0dLRCQ0O1ZcsWrVy5stFrOPPMM7V06VJJ0ocffqiDBw8etU5+fr7CwsLUvn177du3T8uWLZMk9e3bV1lZWUpLS5MkFRQUqLy8XBMnTtQ//vGPqmBY2T0yOTlZa9askSS99tprddZ06NAhJSYmys/PTwsWLKgalGXixImaM2eOioqKqu03ODhY5513nmbOnKnp06ef9DU5GYQ2AAAAoBWJjY3VGWecoQEDBuiee+45avn555+v8vJyDRo0SL/97W81atSoRq/hoYce0ocffqhhw4Zp2bJlSkxMVERERLV1Bg8erKFDhyolJUUzZszQGWecIUkKCgrSkiVLdPvtt2vw4MGaMGGCSkpKdP3116tbt24aNGiQBg8erEWLFlUd6xe/+IXOOuss+fv711nTLbfconnz5mnUqFH67rvvqlrhzj//fE2aNEmpqakaMmSIZs2aVbXNVVddJWOMJk6c2NiX6LiYhjRdNoXU1FS7evVqnxwbAAAAaCqbN29Wv379fF2GTx05ckT+/v4KCAjQihUrNHPmzKqBUVqSWbNm6dChQ3rsscdOel+1vS+MMWustan1bcs9bQAAAAAa1a5du3T55ZfL4/EoKChIzz//vM9q8Vgrj8cqwP/4Ohn+7Gc/07Zt2/Txxx83UWUNR2gDAAAA0Kh69+6tdevWNekxrLUq91iVV3hUVmFV5vGovMKqzP298vlyj0dhQQHqmRB+XPuvHP2yOSC0AQAAAGhWKjzVg1etgcxjax2lMsDPT4H+RoH+fgoJcv7bLqBlD+VBaAMAAABwSnisVxCrcINYzYBW4ZGnljDmb4wC/J1AFtYuQAFuMAv0N25Q81OAv5FfLcP3t3SENgAAAAAn5aiuim5LWM2AVu7Op+bNGKNAPyeABQf6KSI4oKqlrLLVLMDfT/5+rS+MNRShDQAAAECtrLXyWFu9ZayOQGZVS1dFfz8F+hkF+fspNOiHljHvQObvZ2qd3Bo/aNmdOwEAAACcEI+1Ki2vUOGRcoWHhyu74IjWbdmmCyf9TNuyD+vbvQX6Jitf32Tl67t9BfrRuHF6f/l/tTe/RHlFpTpS5pGfkcLbBSg+IkidokL07uJ/qlO4n/p2jNTAzu1193VXKL5dhZLjwtQlOlQdIoMVE9ZOEcGBCgnyV4C/H4GtAWhpAwAAAFqRyq6K1VrGvAbvcJ73qNzzQ8uYx0p7DhUrICJOT82eJ1kd1VUxONBfSbGhSunUvs6uiv/v2b/qhunXKsgd+OO99947JefcWKx1Bjfx82tebVvNqxoAAAAAtXLCmEclZRUqKClTbmGp9ueXaHdesdJzCvX9/sPavCdf1936P3r4iaf0/f7DSj9QqN88+JBmzZqlvbl5uvJnP9bkCWdp8rljtO6zf6tLdKi6x4XJz0j9EyMVXpqrS88do54J4UoI9dOdN0/X+DNG6qbp16j0SImCAvzl72c0c+ZMpaamKiUlRQ899JAk6ZlnnlFWVpbGjRuncePGSZKSk5OVk5MjSXrqqac0YMAADRgwQE8//bQkKT09Xf369dMNN9yglJQUTZw4UcXFxUed+9tvv63TTz9dQ4cO1bnnnqt9+/ZJkg4fPqzp06dr4MCBGjRokF577TVJ0vvvv69hw4Zp8ODBGj9+vCTp4Ycf1qxZs6r2OWDAAKWnp1fVcMstt2jYsGHKyMio9fwkKS0tTWPGjNHgwYM1cuRIFRQU6Kyzzqo2cfgZZ5yhDRs2NNrrLtHSBgAAADSdZfdKe7+udzUrK2slq8r7yNxWH8l53utxcWx/7Rn9Q5Dw96u8R8wovF2Afn7FFXro/nv0qzt/oUB/o+Xvv6X3ly1T584J+uDdtxQZGamcnByNGjVKU6+4tKp7Ys2uin//+98VGhqqDRs2aMOGDRo2bFjVsscff1wxMTGqqKjQ+PHjtWHDBt1xxx166qmn9MknnyguLq7a+a1Zs0Zz587Vl19+KWutTj/9dJ1zzjmKjo7W1q1btXjxYj3//PO6/PLL9dprr+nqq6+utv2ZZ56plStXyhijF154QU8++aT+9Kc/6bHHHlP79u319dfONT548KCys7N1ww036NNPP1X37t2Vm5tb7/X/9ttvNXfuXD333HN1nl/fvn01ZcoULVmyRCNGjFB+fr5CQkJ0/fXX68UXX9TTTz+t7777TkeOHNGgQYPqPebxILQBAAAATcRW/q8yeLnBzFP1uHLZ0dsaIxk5oyv6+Zmqx4HBAQqJD3e6Lfr5ya9GV8WuZ4/SHQdyVHgwW9nZ2YqJjlZSUpLKysp0//3369NPP5Wfn592796tffv2qWPHjrXW/umnn+qOO+6QJA0aNKhaEFm6dKlmz56t8vJy7dmzR5s2bTpmUPn888/1s5/9TGFhYZKkyZMn67PPPtOkSZPUvXt3DRkyRJI0fPhwpaenH7V9ZmampkyZoj179qi0tFTdu3eXJH300Ud6+eWXq9aLjo7W22+/rbPPPrtqnZiYmDrrqpSUlKRRo0Yd8/yMMUpMTNSIESMkSZGRkZKkyy67TI899pj+93//V3PmzNG0adPqPd7xIrQBAAAAx8laq/zicu3NL9G+/BLtzS/R/vwS7cs/ovM6l+v7/QXOfWQD75MdWH1bI1XNN1Y5v1jlsPY/jKx47FEV29VT36WXXqpXX31Ve/fu1RVXXCFJWrhwobKzs7VmzRoFBgYqOTlZJSUlx9xPbcffsWOHZs2apbS0NEVHR2vatGn17qe2SbCrzqXdD2fj7+9fa/fI22+/XXfeeacmTZqk5cuX6+GHH67ab80aa3tOkgICAuTxmnLAu+bKMHms86trv6GhoZowYYL+9a9/aenSpVq9enWd53qiCG0AAABN7Eh5hfYeKlFWXomy8oqdn0PO4335JdUGhEDzV1ru0f6CEpWUHT3nWFRooMYnxsvfz0/tAmoMce8VyJp6xMQrrrhCN9xwg3JycvSf//xHknTo0CElJCQoMDBQn3zyiXbu3HnMfZx99tlauHChxo0bp40bN1bdp5Wfn6+wsDC1b99e+/bt07JlyzR27FhJUkREhAoKCo7qHnn22Wdr2rRpuvfee2Wt1RtvvKEFCxY0+HwOHTqkzp07S5LmzZtX9fzEiRP1t7/9reoeuYMHD2r06NG69dZbtWPHjqrukTExMUpOTtY777wjSVq7dq127NhR67HqOr++ffsqKytLaWlpGjFihAoKChQSEqKAgABdf/31uuiii3TWWWc1qGXveBHaAAAAToLHY5VTeER73EC2O69Yew5VD2fZBUeO2i42zBkivUt0qNoFMDZcSxLgb5QQ0U4dIoPVITJYHdsHq0NEsBIi2yk40F+bN29W97iw+nfUhFJSUlRQUKDOnTsrMTFRknTVVVfpoosuUmpqqoYMGaK+ffsecx8zZ87U9OnTNWjQIA0ZMkQjR46UJA0ePFhDhw5VSkqKevTooTPOOKNqmxtvvFEXXHCBEhMT9cknn1Q9P2zYME2bNq1qH9dff72GDh1aa1fI2jz88MO67LLL1LlzZ40aNaoqcP3mN7/RrbfeqgEDBsjf318PPfSQJk+erNmzZ2vy5MnyeDxKSEjQv//9b11yySWaP3++hgwZohEjRqhPnz61Hquu8wsKCtKSJUt0++23q7i4WCEhIfroo48UHh6u4cOHKzIyUtOnT2/Q+Rwvc6ymyqaUmppqm6LpEAAAoDEdPlKuPW4Yy8or0Z5DlY+dcLYnr0SlFdVbXEIC/dUpKlidokLUqX2IOkWFKDEqWJ2j3MftgxUc6O+jM0JT27x5s/r16+frMnAKZWVlaezYsdqyZUud0wXU9r4wxqyx1qbWt39a2gAAQJtVVuHR3kMlVS1jTitZcbVujPkl5dW28TNSx0gnkA3qEqXzBzhhLLF9iDq5wax9SCATBgNtxPz58/XAAw/oqaeearL53QhtAACgVbLWKrewVHsOlVRrGat6nFeifQUlR43aFxUaqE7tQ9QlOkQju8dUtYxVtpIlRLRTgD/dGQE4pk6dqqlTpzbpMQhtAACgRSourVDWoeIf7h2rbB075ASy3XnFOlJevdtiUICfG76CdWbvOHWKClHnqGC3lcx5PjSIr0cAmhf+VQIAAM1OeYVH+wuOuPePlbgtY16PDxXrYFFZtW2MkRIi2qlTVIj6JUZqfL8Et5UspCqoxYQF0W0Rp0Rdw8OjbTrZcUQIbQAA4JSy1upQcVm1ljHve8j2HHLmvKqoMQx+RHBAVRfFod2i3FYyp+tip6gQdYgMVhCjMKIZCA4O1oEDBxQbG0twg6y1OnDggIKDg094H4Q2AADQqErKKuckqz78vffjotKKatsE+fupY/tgdYoK1unufWTeIy4mtg9WRHCgj84IOD5dunRRZmamsrOzfV0Kmong4GB16dLlhLcntAEAgAbzeKxyDh+pc/j7rLxi5RwuPWq7uPB26hwVrN4J4Tq7d3zVKIuJbrfFuLB28vOjRQKtQ2BgoLp37+7rMtCKENoAAECV/JKyGpNEVx/gY++hEpVVVO+2GBbkX9UyltIp8qh5yTq2D1a7AOYkA4ATRWgDAKCNKC33aF9+3cPfZ+UVq+BI9TnJAvyMOkQ64Wt4t2i3Zaz6iIuRwQHctwMATYjQBgBAK1FUWq7t2YXKPFhU6yAf2YePHDUnWWxYkBKjgpUUG6rRPWPVKSq42oiL8RHt5E+3RQDwKUIbAAAtiMdjtSe/RNv2H9b27MPanlOo7dmF2pZ9WHsOlVRbNyTQv6qL4rjTEpToBrLKgT0S24coJIhuiwDQ3BHaAABohgqPOK1m23MOa5sbyrZnF2pHzmGVlP0wYXREuwD1SAjX6B6x6hEfph7x4eoWE6rOUSGKCg2k2yL/g194AAAgAElEQVQAtAKENgAAfMTjsdqdV6ztOYVOy1mOE8y2Zxdqb/4PrWZ+RuoaE6oecWEa0zNWPePD3YAWpvjwdgQzAGjlCG0AADSxgpKyqlazylC2LfuwduQU6kj5D61mkcEB6hEfrjN6xalHfJh6ui1nSbGhjL4IAG0YoQ0AgEZQ4bHafbBY26qC2eGqLo37C45UrefvZ9Q1OkQ948N1Vu849YgPV4+4MPVMCFdsWBCtZgCAoxDaAAA4DvmVrWZeoWx7dqF2HChUqVerWfuQQPWMD9PZfeLdVrNw9YwPU7eYMAUF+PnwDAAALQ2hDQCAGio8VpkHi6pC2baqkFaonMPVW82SYkLVIz5M55wWX9WdsUdcmGJoNQMANBJCGwCgzTpUVFZrd8adB4pUWvFDq1l0aKB6xIfrR33jq0JZ5SiNtJoBAJoaoQ0A0KqVV3iUcbDYmdPMa+j87TmHlXO4tGq9AD+jpNhQJ5z1S1DPuHD1TAhTj7hwRYcF+fAMAABtHaENANAq5BWVVpvPrHLi6Z0HClVWYavWiw0LUo/4MI3v26EqlPWID1PXmFAF+tNqBgBofghtAIAWo6zCo125RT+EssqQllOo3MIfWs0C/Y2SYsPUIy5ME/p3qOrO2DM+TFGhtJoBAFoWQhsAoNnJLSytFsq2ud0Zdx0oUrnnh1azuPAg9YgP13kpHapazHrGh6tLdIgCaDUDALQShDYAgE+UljutZjW7M27PPqyDRWVV6wX5+yk5LlR9EiJ0fkrHqhazHvHhah8S6MMzAADg1CC0AQCajLVWuYWlVUPmV4aybdmF2pVbpAqvVrP4iHbqERem8wckqqfbYtYjPkxdokPl78fQ+QCAtovQBgA4aaXlHu08UFjVjXHbfue/27MLdajYq9UswE/dY8PUt2OEfjwwUT0q5zWLD1NkMK1mAADUhtAGAGgQa61yDpdWtZR5t5ztyi2SV6OZEiLaqWd8uH4yKLEqlPWKD1enqBBazQAAOE6ENgBArcoqPPrgm736eMv+qpBWUFJetbxdgJ+6x4UppVN7XTS4U1V3xu5xYYqg1QwAgEZDaAMAVLMvv0SLvtylxat2aX/BEcWFB6lPhwhdPKQymIWrR1yYOkeFyI9WMwAAmhyhDQAga61W7cjV/JU79cHGvaqwVuf0idcfRyfrnD7xhDMAAHyI0AYAbVjhkXK9+dVuLVixU1v2FigyOEDTz0jW1aOSlBQb5uvyAACACG0A0CZtzz6sBSt36tU1mSooKVf/xEj98ZKBmjS4s0KC/H1dHgAA8EJoA4A2osJj9fGW/Zq/Il2fbc1RoL/RhQMTNXV0koZ1i5YxdIEEAKA5IrQBQCuXW1iqJWkZemnlTu3OK1bHyGDdNaGPrhjZTfER7XxdHgAAqAehDQBaqfUZeZq3Il3vbNij0nKPRveI1W9+3E8T+ndQgL+fr8sDAAANRGgDgFakpKxC727Yo/kr0rU+85DCgvw1JbWrrhmdpD4dInxdHgAAOAGENgBoBTJyi7Twy11aujpDuYWl6hkfpkcmpWjysM5MdA0AQAtHaAOAFsrjsfr8+xzNX7FTH2/ZJ0ma0L+Drh2drNE9YxlYBACAVoLQBgAtTH5JmV5dnamXVu7U9pxCxYYF6ZaxvXTl6d3UKSrE1+UBAIBGRmgDgBZiy958zV+xU2+u262i0goN6xalp6cM0QUDO6pdAHOrAQDQWhHaAKAZK6vw6INv9mr+ip1atSNX7QL8dPGQTpo6OlkDOrf3dXkAAOAUILQBQDO0P79Ei1bt0qIvd2l/wRF1jQnR/Rf21WXDuyo6LMjX5QEAgFOI0AYAzYS1VmnpBzVvRbo+2LhX5R6rsafF64nRSTqnT4L8/RhYBACAtojQBgA+VlRarjfXZWn+inRt2VugyOAATRuTrKtHJSk5LszX5QEAAB8jtAGAj2zPPqwFK3fq1TWZKigpV//ESD0xeaAuHtJZIUEMLAIAABwNCm3GmPMl/UWSv6QXrLVP1FjeTdI8SVHuOvdaa99r5FoBoMWr8Fh9vGW/5q9I12dbcxTob3TBgERdOyZJw7pFM7caAAA4Sr2hzRjjL+lZSRMkZUpKM8a8Za3d5LXabyQttdb+3RjTX9J7kpKboF4AaJFyC0u1JC1DL63cqd15xeoYGay7JvTRlJFdlRAR7OvyAABAM9aQlraRkr631m6XJGPMy5IuluQd2qykSPdxe0lZjVkkALRU6zPyNH/FTr29IUul5R6N6hGj3/y4n87t30GB/n6+Lg8AALQADQltnSVleP2eKen0Gus8LOlDY8ztksIkndso1QFAC1RSVqF3N+zR/JU7tT4jT2FB/pqS2lXXjE5Snw4Rvi4PAAC0MA0JbbXdYGFr/P5zSS9aa/9kjBktaYExZoC11lNtR8bcKOlGSerWrduJ1AsAzVbmwSIt/HKXlqRlKLewVD3jw/TIpBRNHtZZEcGBvi4PAAC0UA0JbZmSunr93kVHd3+8TtL5kmStXWGMCZYUJ2m/90rW2tmSZktSampqzeAHAC2Ox2P13205mvfFTn28ZZ8kaUL/Dpo6OlljesYysAgAADhpDQltaZJ6G2O6S9ot6QpJV9ZYZ5ek8ZJeNMb0kxQsKbsxCwWA5iS/pEyvrcnUghU7tT2nULFhQZo5tqeuPD1JnaNCfF0eAABoReoNbdbacmPMbZI+kDOc/xxr7TfGmEclrbbWviXpLknPG2P+R07XyWnWWlrSALQ6W/bma/6KnXpz3W4VlVZoaLcoPT1liC4Y2FHtAphbDQAANL4GzdPmzrn2Xo3nHvR6vEnSGY1bGgA0D2UVHn3wzV7NX7FTq3bkql2AnyYN7qSpo5M1sEt7X5cHAABauQaFNgBoi/bnl2jRql1avGqX9uUfUdeYEN1/YV9dNryrosOCfF0eAABoIwhtAODFWqu09IOavyJd72/cq3KP1djT4vWHyUk6p0+C/P0YWAQAAJxahDYAkFRUWq4312Vp/op0bdlboMjgAE0bk6yrRyUpOS7M1+UBAIA2jNAGoE3bkVOoBSt26pU1GSooKVf/xEg9MXmgLh7SWSFBDCwCAAB8j9AGoM2p8Fh9smW/5q1I12dbcxTob3TBgERNHZ2k4UnRzK0GAACaFUIbgDbjYGGplqzO0EsrdyrzYLE6Rgbrzgl9dMXIrkqICPZ1eQAAALUitAFo9TZk5mneFzv19oYslZZ7NKpHjB64sJ/O7d9Bgf5+vi4PAADgmAhtAFqlkrIKvbthj+av3Kn1GXkKC/LXlNSuumZ0kvp0iPB1eQAAAA1GaAPQqmQeLNLCL3dpSVqGcgtL1TM+TI9MStHkYZ0VERzo6/IAAACOG6ENQItnrdXn3+do/oqd+r/N+yRJE/p30NTRyRrTM5aBRQAAQItGaAPQYuWXlOm1NZlasHKntmcXKjYsSDPH9tSVpyepc1SIr8sDAABoFIQ2AC3Ot3sLNH9Fut5Yt1tFpRUa2i1Kf54yWBcOTFS7AOZWAwAArQuhDUCLUFbh0Yff7NO8FelatSNX7QL8NGlwJ00dnayBXdr7ujwAAIAmQ2gD0Kztzy/R4lUZWrRqp/blH1HXmBDdf2FfXTa8q6LDgnxdHgAAQJMjtAFodqy1Wr3zoOZ9ka73N+5VucfqnD7x+sPkJJ3TJ0H+fgwsAgAA2g5CG4Bmo6i0XG+uy9L8FenasrdAkcEBmjYmWVePSlJyXJivywMAAPAJQhsAn9uRU6gFK3bqlTUZKigpV7/ESD0xeaAuHtJZIUEMLAIAANo2QhsAn6jwWH2yZb/mr9ypT7/LVqC/0QUDEjV1dJKGJ0UztxoAAICL0AbglDpYWKolqzP00sqdyjxYrI6RwbpzQh9dMbKrEiKCfV0eAABAs0NoA3DKvLMhS3e/sl4lZR6N6hGj+y/spwn9OyjQ38/XpQEAADRbhDYAp8SL/92hR97ZpNSkaD3+s4Hq0yHC1yUBAAC0CIQ2AE3KWqsnP/hWf1++TeeldNBfrhiq4EAGFwEAAGgoQhuAJlNW4dGvX9ug19fu1tWjuumRSQOYYw0AAOA4EdoANInCI+W6ZeFa/ee7bN09sY9uHdeLESEBAABOAKENQKM7cPiIZryYpo1Z+frjJQM1ZUQ3X5cEAADQYhHaADSqXQeKNHXOl9qbX6LZ1wzX+H4dfF0SAABAi0ZoA9BoNu4+pGlz01Tu8WjRDaM0rFu0r0sCAABo8QhtABrF51tzdNOC1YoKDdLLM0apV0K4r0sCAABoFQhtAE7av77arbtfWa+e8eGaN2OkOkQG+7okAACAVoPQBuCkvPDZdv3u3c0a1SNGs6emKjI40NclAQAAtCqENgAnxOOx+sOyzXr+sx368cBEPTVlsNoFMGk2AABAYyO0AThupeUe3fPqev3rqyxNG5OsB3/SX35Mmg0AANAkCG0AjsvhI+W6ecEaff59jn59fl/dfE4PJs0GAABoQoQ2AA2WXXBE019cpc17CjTrssG6dHgXX5cEAADQ6hHaADTIjpxCXTtnlbILjuiFa1M17rQEX5cEAADQJhDaANRrfUaeZryYJitp8Y2jNKRrlK9LAgAAaDMIbQCOafm3+3XLwrWKDQ/S/Bmnq3tcmK9LAgAAaFMIbQDq9PraTP3q1Q06rWOE5k4foYQIJs0GAAA41QhtAI5irdX/+3S7nli2RWf0itU/rh6uCCbNBgAA8AlCG4BqPB6rx97dpLn/TdekwZ0067LBCgrw83VZAAAAbRahDUCVI+UVunPper27YY+uO7O7HriwH5NmAwAA+BihDYAkKb+kTDfNX6MV2w/ogQv76Yaze/i6JAAAAIjQBkDS/vwSXTs3TVv3FejpKUP006GdfV0SAAAAXIQ2oI3bln1YU/+5SnlFpZozbYTO7hPv65IAAADghdAGtGHrdh3UjBfT5O9n9PKNozWwS3tflwQAAIAaCG1AG/Xxln26ZeFadYgM1vwZI5UUy6TZAAAAzRGhDWiDlqZl6L43vlb/xEjNnT5CceHtfF0SAAAA6kBoA9oQa62e/eR7zfrwO53VO07/uHq4wtrxzwAAAEBzxrc1oI2o8Fg9/NY3WrByp342tLP+eMkgJs0GAABoAQhtQBtQUlahX778ld7/Zq9uOqeHfn1eXybNBgAAaCEIbUArd6i4TDfMX61VO3L125/013Vndvd1SQAAADgOhDagFdtzqFjT5qRpe85hPfPzoZo0uJOvSwIAAMBxIrQBrdTWfQW6ds4q5ZeUa970kRrTK87XJQEAAOAEENqAVmh1eq6um7daQQF+WnLTKKV0YtJsAACAlorQBrQyH36zV7cvXqdOUSGaP2OkusaE+rokAAAAnARCG9CKLPpyl37z5tca2CVKc65NVSyTZgMAALR4hDagFbDW6i//t1VPf7RV406L17NXDVNoEB9vAACA1oBvdUALV17h0W//9Y0Wr9qlS4d30R8mD1SgP5NmAwAAtBaENqAFKymr0O2L1+nfm/bptnG9dNfEPjKGSbMBAABaE0Ib0ELlFZXqunmrtXbXQT16cYqmjk72dUkAAABoAoQ2oAXanVesa+es0q4DRXr2ymG6cGCir0sCAABAEyG0AS3Mt3udSbMLS8s1/7qRGtUj1tclAQAAoAkR2oAW5MvtB3T9/NUKDfLXKzePVt+Okb4uCQAAAE2M0Aa0EMu+3qNfLPlKXaNDNP+609U5KsTXJQEAAOAUILQBLcCCFel68K1vNLRrlP557QhFhwX5uiQAAACcIoQ2oBmz1upPH36nv33yvc7t10F//flQhQT5+7osAAAAnEKENqCZKq/w6P43vtbS1Zn6+ciueuziAQpg0mwAAIA2h9AGNENFpeW6bdE6fbxlv34xvrd+eW5vJs0GAABoowhtQDOTW1iqGS+maUNmnh7/2QBddXqSr0sCAACADxHagGYkI7dI185Zpd15xfr71cN1XkpHX5cEAAAAHyO0Ac3Epqx8XTt3lY6UVeil60/XiOQYX5cEAACAZoDQBjQDX2zL0U3z1yg8OEALZ45Rnw4Rvi4JAAAAzQShDfCxdzZk6c4l65UcF6p5M0YqsT2TZgMAAOAHhDbAh+b+d4cefWeTRiTF6PmpqWofGujrkgAAANDMENoAH7DW6o/vf6t//GebzkvpoL9cMVTBgUyaDQAAgKM1aKZeY8z5xphvjTHfG2PurWOdy40xm4wx3xhjFjVumUDrUVbh0V2vrNc//rNNV4/qpueuGk5gAwAAQJ3qbWkzxvhLelbSBEmZktKMMW9Zazd5rdNb0n2SzrDWHjTGJDRVwUBLVnikXDMXrtWn32Xr7ol9dOu4XkyaDQAAgGNqSPfIkZK+t9ZulyRjzMuSLpa0yWudGyQ9a609KEnW2v2NXSjQ0uUcPqIZL6bpm6x8/fGSgZoyopuvSwIAAEAL0JDQ1llShtfvmZJOr7FOH0kyxvxXkr+kh6217zdKhUArsOtAkabO+VJ780s0+5rhGt+vg69LAgAAQAvRkNBWW98tW8t+eksaK6mLpM+MMQOstXnVdmTMjZJulKRu3WhlQNuwcfchTZu7SuUeq0U3jNKwbtG+LgkAAAAtSEMGIsmU1NXr9y6SsmpZ51/W2jJr7Q5J38oJcdVYa2dba1Ottanx8fEnWjPQYny2NVtT/t8KtQvw16s3jyGwAQAA4Lg1JLSlSeptjOlujAmSdIWkt2qs86akcZJkjImT011ye2MWCrQ0b67brelz09Q1JlSv3zJGvRLCfV0SAAAAWqB6u0daa8uNMbdJ+kDO/WpzrLXfGGMelbTaWvuWu2yiMWaTpApJ91hrDzRl4UBz9vyn2/X4e5s1qkeMZk9NVWQwk2YDAADgxBhra96edmqkpqba1atX++TYQFPxeKx+/95mvfD5Dv14YKKemjJY7QKYgw0AAABHM8assdam1rdeQwYiAdAApeUe3f3Ker21PkvTxiTrwZ/0l58fc7ABAADg5BDagEZQUFKmmS+t1eff5+jX5/fVzef0YNJsAAAANApCG3CS9heUaPrcNG3ZW6BZlw3WpcO7+LokAAAAtCKENuAk7Mgp1NQ5XyqnoFQvXJuqcacl+LokAAAAtDKENuAErc/I0/QX0yRJi28cpSFdo3xcEQAAAFojQhtwApZ/u18zX1qruIggzZ9xurrHhfm6JAAAALRShDbgOL22JlO/fm2DTusYobnTRyghItjXJQEAAKAVI7QBDWSt1T/+s11/fH+LzugVq39cPVwRTJoNAACAJkZoAxrA47F69J1NevGLdE0a3EmzLhusoAA/X5cFAACANoDQBtTjSHmF7ly6Xu9u2KPrz+yu+y/sx6TZAAAAOGUIbcAx5JeU6cb5q7Vye64euLCfbji7h69LAgAAQBtDaAPqsC+/RNfOWaVt2Yf19JQh+unQzr4uCQAAAG0QoQ2oxbbsw5r6z1XKKyrVnGkjdFbveF+XBAAAgDaK0AbUsHbXQV33Ypr8/YyW3DRaAzq393VJAAAAaMMIbYCX/9u8T7cuWquOkcGaN2OkkmKZNBsAAAC+RWgDXEvTMnTfG18rpVOk5kwbobjwdr4uCQAAACC0AdZa/e3j7/Wnf3+ns/vE6+9XDVNYOz4aAAAAaB74Zoo2rcJj9dBbG/XSyl2aPLSz/njpIAX6M2k2AAAAmg9CG9qskrIK/fLlr/T+N3t18zk99evzT5MxTJoNAACA5oXQhjbpUFGZbpi/Wmk7c/XgT/prxpndfV0SAAAAUCtCG9qcPYeKde2cVUrPKdIzVwzVRYM7+bokAAAAoE6ENrQpW/cVaOqcVSooKdeL00doTK84X5cEAAAAHBOhDW3G6vRcXTdvtYIC/LTkplFK6cSk2QAAAGj+CG1oEz74Zq/uWLxOnaNCNG/GSHWNCfV1SQAAAECDENrQ6i38cqd+++ZGDeoSpTnTRigmLMjXJQEAAAANRmhDq2Wt1dMfbdVf/m+rxp0Wr2evGqbQIN7yAAAAaFn4BotWqbzCo9/+a6MWr8rQZcO76PeTBzJpNgAAAFokQhtaneLSCt2+eJ0+2rxPt43rpbsm9mHSbAAAALRYhDa0KnlFpbpu3mqt3XVQj12comtGJ/u6JAAAAOCkENrQauzOcybN3pVbpOeuHKYLBib6uiQAAADgpBHa0Cps2Zuva+esUlFphRbMGKnTe8T6uiQAAACgURDa0OKt3H5AN8xfrbCgAL1y82j17Rjp65IAAACARkNoQ4v2/sY9uuPlr9QtJlTzZoxU56gQX5cEAAAANCpCG1qsVTtydduidRrUpb3mTBuhqFAmzQYAAEDrQ2hDi7TnULFuWbhG3WJC9eKMkYoMDvR1SQAAAECTILShxSkpq9DNC9aopMyjl28cTmADAABAq0ZoQ4tirdVv39yo9ZmHNPua4eqVEOHrkgAAAIAm5efrAoDjsWDlTr2yJlN3jO+tiSkdfV0OAAAA0OQIbWgxvtx+QI++vUnn9kvQL8f39nU5AAAAwClBaEOLkJVXrFsXrVW32FA9NWWI/PyMr0sCAAAATgnuaUOzV1JWoZkvVQ48ksrAIwAAAGhTCG1o1qy1+o078MjzU1PVKyHc1yUBAAAApxTdI9GszV+xU6+uydQvxvfWhP4dfF0OAAAAcMoR2tBsfbn9gB57Z5PO7ddBv2DgEQAAALRRhDY0S1l5xbploTPwyJ+nDGbgEQAAALRZhDY0OyVlFbr5pTU6Uu7R7GtSFcHAIwAAAGjDGIgEzYq1Vg+8sVEbGHgEAAAAkERLG5qZeV+k67W1DDwCAAAAVCK0odlYuf2AHnt3MwOPAAAAAF4IbWgWducV69aFa5XEwCMAAABANYQ2+FxJWYVuXsDAIwAAAEBtGIgEPmWt1f1vfK2vdzPwCAAAAFAbWtrgUy9+ka7X1+7WL89l4BEAAACgNoQ2+MyKbQf0u3c3a0L/DrrjRww8AgAAANSG0Aaf2J1XrFsXrVVybKieupyBRwAAAIC6ENpwypWUVeimBatVVu7R7KkMPAIAAAAcCwOR4JSy1ur+17/Wxt35emFqqnrGM/AIAAAAcCy0tOGUmvvfdL2+brf+59w+OpeBRwAAAIB6EdpwynyxLUePv7dZE/t30O0/6uXrcgAAAIAWgdCGUyLzYJFuW7ROybGh+hMDjwAAAAANRmhDkyspq9DNL61h4BEAAADgBDAQCZqUtVb3vf61vsli4BEAAADgRNDShiY157/pesMdeGR8PwYeAQAAAI4XoQ1N5ottOfq9O/DIbeMYeAQAAAA4EYQ2NInKgUe6x4XpqSlDGHgEAAAAOEGENjS64tIK3bRgjcoqPJp9zXCFt+PWSQAAAOBE8W0ajcoZeGSDNu3J1z+vTVUPBh4BAAAATgotbWhU//x8h978Kkt3nttHP+rLwCMAAADAySK0odF88X2O/rBsi85L6aBbGXgEAAAAaBSENjSKjNwi3bporXrEhelPlzPwCAAAANBYCG04aZUDj5R7rGZPTWXgEQAAAKAR8e0aJ8Vaq3tf36DNe/M159oR6h4X5uuSAAAAgFaFljaclH9+vkP/+ipLd03oo3F9E3xdDgAAANDqENpwwv77fY5+/95mnZ/SkYFHAAAAgCbSoNBmjDnfGPOtMeZ7Y8y9x1jvUmOMNcakNl6JaI4ycot026K16hkfrlmXD5YxDDwCAAAANIV6Q5sxxl/Ss5IukNRf0s+NMf1rWS9C0h2SvmzsItG8MPAIAAAAcOo0pKVtpKTvrbXbrbWlkl6WdHEt6z0m6UlJJY1YH5oZa61+/Zoz8MgzPx/KwCMAAABAE2tIaOssKcPr90z3uSrGmKGSulpr32nE2tAMvfDZDr21Pkt3TzxN405j4BEAAACgqTUktNV2s5KtWmiMn6Q/S7qr3h0Zc6MxZrUxZnV2dnbDq0Sz8PnWHP1h2WZdMKCjbhnb09flAAAAAG1CQ0JbpqSuXr93kZTl9XuEpAGSlhtj0iWNkvRWbYORWGtnW2tTrbWp8fHxJ141TrmM3CLdtniteiWEa9ZlDDwCAAAAnCoNCW1pknobY7obY4IkXSHprcqF1tpD1to4a22ytTZZ0kpJk6y1q5ukYpxyxaUVunHBGnk8VrOvSVUYA48AAAAAp0y9oc1aWy7pNkkfSNosaam19htjzKPGmElNXSB8y1qrX722QVvcgUeSGXgEAAAAOKUa1GRirX1P0ns1nnuwjnXHnnxZaC6e/2y73l6fpXvOO01jGXgEAAAAOOUaNLk22qbPtmbriWVbdOFABh4BAAAAfIXQhlpl5Bbp9sXr1DshQv97KQOPAAAAAL5CaMNRikrLdcP81c7AI1OHM/AIAAAA4EN8G0c11lr96tUN+nZfgeZOG6GkWAYeAQAAAHyJlrb/3969h1dV3/kef/+ScIdwExCDyEUFRW0VKlZr1Tq22lqltp7qqDOtOrQ8tbY6HTvVPr1Op9cz1lY7ai+n1bFeO0W8VATR2uNxqqBYAUFQvCBCuCiEW0KS3/lj7bj3zh1MWHtnv1/Ps57srKyd/U2WxPXZv9/6/pTn5sdf5v6/vWnjEUmSJKlAGNr0jr+s3MAPH1rOx44czayTbDwiSZIkFQJDmwB4bdMOLvv9sxw6ahA/+tRRNh6RJEmSCoShTeyoq2fmrQsBuOkiG49IkiRJhcTQVuJijPzLPX/jxfU1/Oz8o208IkmSJBUYQ1uJu+nxl3ngb2/yLx+ZzEmHjki7HEmSJEnNGNpK2OMvbuBHDy3nY0eN5vMnTUi7HEmSJEmtMLSVqFc3beeLtyeNR35s4xFJkiSpYBnaStCOuno+d+siAG6+aBr9e9t4RJIkSSpUhrYSk9t45OfnH83Y4f3TLkmSJElSOwxtJebGPyeNR646fTIftPGIJEmSVPAMbSXkzy9u4Edzl3PmUaP53AdtPCJJkiQVA0NbiXh103a++PtnmDRqED+y8VzdWFwAAB7kSURBVIgkSZJUNAxtJWB7bT0zb1lECMHGI5IkSVKRMbT1cEnjkedYWV3D9X9v4xFJkiSp2Bjaerj//PNLPPj8Or56+mROPMTGI5IkSVKxMbT1YI+tqObHc1fw8fccwEwbj0iSJElFydDWQ72ycTuX3/4sk0YN4oefPNLGI5IkSVKRMrT1QNtr65l560LKygK//Acbj0iSJEnFzNDWwzQ1HllVvY3rzz+GA4fZeESSJEkqZoa2Hqap8ci/njGZDxyyX9rlSJIkSXqXDG09SG7jkX860cYjkiRJUk9gaOshbDwiSZIk9UyGth7AxiOSJElSz2VoK3I2HpEkSZJ6NkNbkfvFYzYekSRJknoyQ1sRe3RFNT95eAVn2XhEkiRJ6rEMbUXqlY3b+dLtzzJ5/0p++MmjbDwiSZIk9VCGtiK0LafxyM0XTaVf7/K0S5IkSZLUTQxtRSbGyL/cbeMRSZIkqVQY2orMLx57iT8tWcfXzjjMxiOSJElSCTC0FZFHlyeNR85+7wFceuL4tMuRJEmStA8Y2orE6o3bufyOZzls/0p+cI6NRyRJkqRSYWgrAttq65l5y0IqygI32XhEkiRJKikVaReg9sUY+cpdz/HShm3cesl0G49IkiRJJcaRtgJ3w6OreGjpOq7+6GGccLCNRyRJkqRSY2grYAuWr+d/z3uRGe89gEs+YOMRSZIkqRQZ2grU6o3b+dIdizls/0q+b+MRSZIkqWQZ2gqQjUckSZIkNbERSYFpbIz8812LeXnjdm69+Fgbj0iSJEklzpG2AnPDo6uYu3Q9XztjMsfbeESSJEkqeYa2ArJg+Xr+Y76NRyRJkiRlGdoKxMsbtvGl2xdz+Ggbj0iSJEnKMrQVgJpdu5l56yJ6VZTZeESSJElSHhuRpCxpPPIcqzdu59ZLjmXMUBuPSJIkScpypC1l1z+6ioeXrefqjx7G8RNtPCJJkiQpn6EtRY+8sJ5r57/IJ46u4uITxqVdjiRJkqQCZGhLyUsbtvHlOxYz5YBKvn/OkTYekSRJktQqQ1sKanbtZuYtCzONR6bRt5eNRyRJkiS1zkYk+1hjY+TKu57jlU07+K9LplM1pF/aJUmSJEkqYI607WM/X7CKecvWc81HD+P9E4enXY4kSZKkAmdo24fmL0saj5xzTBWftfGIJEmSpE4wtO0jL23YxhV3LubIqsH8+ydsPCJJkiSpcwxt+0BT45HeFWXceNFUG49IkiRJ6jQbkXSzxsbIFXfaeESSJEnS3nGkrZv9bMFK5r+wnq9/zMYjkiRJkvacoa0bzVu2np/OX8k5x1TxmePHpV2OJEmSpCLk9MhusqraxiOSJElSt6uvhdoa2LUl+VhbA7VbM/u2Jo8H7AdTP5N2pXvN0NYNtu7azcxbF9KnooybbDwiSZIktdRQnw1XeSGr6fPm+3L25+5rqO34tca8z9CmrMbGyJV3Lua1TTv4r0unc4CNRyRJXamxEeq2tbyAyb2w6TcERk2BEZOhl/8fktTFGhuhrqYTISvnY4vgVQO7d3T8WqEc+gyCvpXQpzJ5PHB/GH5Izv5B0Gdw5mPuvpzn9Orb/b+XbmRo62LXPbKS+S9U862PH85xE2w8IknKiDG5QKnNvdBp7R3mtt51ztlH7NxrhjIYNjEJcE3byMNhyEFQ5m3tUsmJEeq2txOocv7evDPVcGuzv1s1SWDrUMgJTpkw1X8YDD0oZ19lTsDKCVm5+3r1B28zMrR1pYeXruO6R1byyWPG8I82HpGknqO+NidQtXIBU7ullX2tXBDFho5fq1f/Zhcwg2DQqPyLnLbeSW7atm+E9UugehmsXwpvPgfLZmdfo/fAJLyNOhxGHZF93G9o9/0OJe29GKF+V9tv7uTtb+1vVM7+2Njx6/Ua0DJMDRqdP9rV0d+i3gN9c6gLhRg7+W5dF5s2bVpcuHBhKq/dHVZVb2PGDU8wYcQA7vrc+72PTZIKQcPudqbmtDFdp7ULooa6jl+rvE+zqTnNLmBau7Bp8Q7zICjv1T2/i9ptsGF5EubWL4X1y5LHu97OHlNZlR2NG3VE8ni/Q7qvJqkU1Nd1MH2wk/dtNe7u+LUq+nYcptr9W5T5WOZ17L4SQlgUY5zW0XGOtHWBrbt2M/OWhfTtVcaNF9p4RJLetcaGlmGq3ek6bby7XL+z49cK5S3viag8oGWYypuyk3sBlHlORZ/u/728G30GwphpydYkRqh5MxPicraXHs1eIJb1ghGTMkEuZ5rloNFOWVJpadgNb78Gm1fDltfb6FTYyr76XR1/77JeLcNU5RgY2c7fnLzRrsHJyFZF7+7/PSgVhrZ3qbExcsUdi3lt8w5us/GIpFLX2Ai7t3fyneQtbd+zVbetEy8WWr5L3H84DB3f8qb19t557tWvdMNHCElArTwADjktu7++DjatzI7GVS+DV5+A5+/KHtN3SHY0rmma5YjJSTiUitXuXfD2q7D55Zbb26+3nOIcylrehzVwJAw/uJOjWpXZN31K9e+QOsXQ9i799JGVPLK8mm+fNYXpNh6RVKxihN07O5jC094N6zn7OtMko/fAlu8eD67KH+3q6IKn1wDvl+guFb2zI2qcm92/861MkFsK1ZlRucW35YfsoeObNT6ZAsPGO91KhaNuB7y1upVgthq2rCHvb1ifwTB8AlRNhSPPhWETkm3wgUmXVptkaB8xtL0Lc5eu42ePrORTU8fwD+8/KO1yJJWqFouKtjZ1sK1uhDlTeRrrO36tin4t78UaMKKNDmBtTOPxfoni1W8ojDsh2Zo0NiYjE01NT9YvSYLdigezDQ8q+sHIydkQ1xToBuyXzs+hnm/X1laCWebzmjfzj+0/PAliBx2fDWVNW7+hhjIVBBuR7KVV1TWcff0THDxyIHfaeETS3shtktFemOpotKszTTLeuV+io/uzWmmg4f0S2hu7d2Yan+Q0PVm/FHZszB4zcFSzxieHw36Tin49Je0jOzZng9jml/ND2vYN+ccO3D8njI3Pfhw6Phkxk1JiI5JulDQeWUS/3uXceJGNR6SS09iQTAfr9NTB1m5W39r5JhltLSqaN7I1uI1uhD1jUVEVoV794ICjky3XtursaFzTNMunfgkNtcnXQ3lyP1DuFMtRU5LpaI54lJYYk+UrWru/bPPL+Z1PIWncMWw8TPpo/mjZ0HHea6miZ2jbQ7mNR37/T8cxerCNR6Si8c6iop1ZyLidtW463SRjTxYVbWNfKTfJUM80cCQM/BBM/FB2X0N9chHeNBpXvQzeWAhL/zt7TJ/KnLXlmqZZHp6MAqt4xQg169oIZqvzF3EOZUl4HzYBjvhks2B2UPL3UuqhDG176KfzX+SR5dV85+wpHDt+WNrlSKWhaVHRDhcybmutm8y+un20qGjfSptkSHuivAJGHJpsR5yT3b9rK1S/kG16sn4pPP8HWPib7DGDx+YEucw0y+EHJ99ThaGxEba+0Xooe2s17N6RPbasAoYclASxse/PD2ZDxjpFWyWrU3/RQginA9cB5cCvYow/aPb1K4FLgXpgA3BxjPHVLq41dQ8tWcfPFqzi3KljuOg4G49IeyxGeGNRMj2qvTVtWlt7q1NNMlpZVHTAhFbu2XJRUako9K2EsdOTrUmMSYe/6mX50yxXzc/+nSjvnVlbrtkUy4GjHLnuLg31sOW1/IYf79xr9mp2+iskC9EPHZcEsQkn59xjlunKaOCWWuiwEUkIoRx4ETgNWAM8DZwfY1yWc8wpwF9jjDtCCLOAk2OMn27v+xZbI5KV62uYccMTHDxqEHfOPM772KQ9tXUtzLkcVs1r+bXWFhVt9f4sFxWV1Ib6Wtj4Yv4i4dXL8jsF9h+e3/Rk1BQYcRj07p9e3cWkvq6dNcxey39zrVf/Zk0/crZBBzgTQcroykYkxwKrYowvZ77xHcDZwDuhLcb4aM7x/wNcuGflFrYtO3cz89ZM45ELjzGwSXsiRvjbnfCnq5JuiR/596StsouKSupKFX1g/yOTLdeOzTlBLrNQ+DO/y5mSF5Igkbe23OFJV8FSDBa7d8Jbr7QezLasyZ9i3ntQsobZ6PfAlE/kBzNHNaUu1ZnQVgW8nvP5GmB6G8cCXAL86d0UVUgaGyNX3LmY1208Iu25mvVw/xWw4gE48DiY8QsYPjHtqiSVkv7DYPyJydaksTG5l6ppNG79kmR74T7eWVi514Ds2nKjjsiM0E1Jvl+xq93W9hpmW9/IP7bf0CSEHTgd3nN+fjDrP9xgJu0jnQltrf1rbHVOZQjhQmAacFIbX58JzAQYO3ZsJ0tM17XzX2TB8mq+a+MRac8s+QM88M9QtwM+/D04bpb3ikkqDGVlyRtIwyfC4Wdl99dth+rl+Y1PXrgfnrkle8yg0a2sLXdoMtJXSHa+3TKQvbOGWXX+sQNGJCFs/Afz1zIbOr5nhFSpB+hMaFsDHJjz+RhgbfODQgh/B1wDnBRjrG3+dYAY483AzZDc07bH1e5jDy15k58vWMX/mjaGC208InXO9o1JWFs2G6qmwowbk45wklToeg+AMVOTrUlTS/p3glym8cnqx7ML25dVJMGtaTSuaaus6r6RqBgzi0u3sYbZzs35xw8anYSxQz/crFX++OT+YUkFrTOh7WngkBDCeOAN4Dzg73MPCCEcDdwEnB5jrG75LYrPyvU1/PNdz/GeA4fwnbOPIDj8L3Vs2ZxkOmTtVjj1m3D85XYBk1TcQoDK0cl28N9l9zfshk2r8hufvP5XWHJP9pi+g3M6WGZG5kYeltzL2xkxJt1221rDrHZLbqGZNczGw+Fnt1xc2mYrUlHr8GoqxlgfQrgMmEvS8v83McalIYTvAAtjjHOAHwMDgbsz4ea1GONZbX7TApdtPFLBTRdOtfGI1JEdm5NGI8/fndyQPuO+5AJFknqq8l5JABt5GBz5qez+nW8na8s1NT1ZvxSeuyN/keghB+U3PRkxqZVRs8yUxt3bs88L5claZcMmwJj3tVxcutCmaErqMh22/O8uhdryv6ExcunvnuYvKzdy+8zjeN8453JL7VrxENx3OezYBB+8Ck68MrmYkSQlYkxa4q9fmj/NctPK/G6MkCyB0rSGWd42Pgls/n2VepSubPlfUq6d9yKPrtjAd2ccYWCT2rNrCzz0NVh8WzL954LMKJskKV8IyUjY0INg8kez+3fvgo0rYOPKpBPjsAkweIxNmyS1YGjL8dCSN7n+0UzjkenF0d1SSsWqR2DOF5Ob80/8Cpz0VRe2lqQ91atv8maXb3hJ6oChLcf/e2mTjUek9tTWwMNfh0W/hf0mwaXzkg6RkiRJ6jaGthzfPmsKO+oabDwitWb14zD7C7Dl9aQr5CnXJO8SS5IkqVsZ2nKEEBjQx1+JlKduO8z/Fjx1MwybCBfPhbHT065KkiSpZJhQJLXt1Sdh9ix4azVMnwWnfsO1fiRJkvYxQ5uklnbvhAX/Bk/ekLSY/swDMO4DaVclSZJUkgxtkvKtWQh//HyyftC0S+C070CfgWlXJUmSVLIMbZIS9bXw2Pfhieugsgoumg0TT0m7KkmSpJJnaJMEa5+FP86CDS/A0RfBR/4d+lamXZUkSZIwtEmlrb4O/vITePwnMHAkXHAPHHJa2lVJkiQph6FNKlXrlsDsz8O65+Go8+CMH0C/oWlXJUmSpGYMbVKpaaiHJ66Fx34I/YbAeb+HyR9LuypJkiS1wdAmlZLq5cno2tpn4YhPwhk/hgHD065KkiRJ7TC0SaWgsQGevB4WfC9p33/ub2HKJ9KuSpIkSZ1gaJN6uo2rYPYsWPMUTD4Tzrw2aToiSZKkomBok3qqxkZ46iaY/22o6APn/AqO/BSEkHZlkiRJ2gOGNqkn2rwa7v0CvPoEHPIR+Ph1UDk67aokSZK0FwxtUk/S2AiLfgMPfwPKyuHsG+C9Fzi6JkmSVMQMbVJP8fZrcO9lsPrPMPFDcNbPYfCYtKuSJEnSu2Rok4pdjPDMLTD3GiDCmT+FqZ9xdE2SJKmHMLRJxWzrWphzOayaB+NOTKZDDj0o7aokSZLUhQxtUjGKEf52J/zpKmjYnSyS/b5Loaws7cokSZLUxQxtUrGpWQ/3XwErHoADj4MZv4DhE9OuSpIkSd3E0CYVixhhyR/gwa9A3Q748PfguFlJl0hJkiT1WIY2qRhs3wgPXAnL7oWqqTDjRhhxaNpVSZIkaR8wtEmFbtmcZDpk7VY49Ztw/OVQ7j9dSZKkUuGVn1SodmxOGo08fzeMfg/MuA9GHZ52VZIkSdrHDG1SIVrxENx3OezYBCdfDSdeCeW90q5KkiRJKTC0SYVk59sw92pYfBuMnAIXZEbZJEmSVLIMbVKhWDU/WSi7Zh2c+BU46atQ0TvtqiRJkpQyQ5uUttoaePjrsOi3sN8kuHRe0iFSkiRJwtAmpWv14zD7C7Dl9aQr5CnXQK++aVclSZKkAmJok9JQtx3mfwueuhmGTYSL58LY6WlXJUmSpAJkaJP2tVefhNmz4K3VMH0WnPoN6N0/7aokSZJUoAxt0r6yeycs+Dd48gYYMhY+8wCM+0DaVUmSJKnAGdqkfWHNQvjj52HTSph2CZz2HegzMO2qJEmSVAQMbVJ3qq+Fx74PT1wHgw6Ai2bDxFPSrkqSJElFxNAmdZe1z8IfZ8GGF+Doi+Aj34O+g9OuSpIkSUXG0CZ1tfo6+MtP4PGfwMCRcME9cMhpaVclSZKkImVok7rSuiUw+/Ow7nk46jw44wfQb2jaVUmSJKmIGdqkrtBQD09cC4/9EPoNgfN+D5M/lnZVkiRJ6gEMbdK7Vb08GV1b+yxMOQc++hMYMDztqiRJktRDGNqkvdXYAE9eDwu+B70HwLm/hSmfSLsqSZIk9TCGNmlvbFwFs2fBmqdg8plw5rVJ0xFJkiSpixnapD3R2Ah/vREe+TZU9IVzfglHngshpF2ZJEmSeihDm9RZm1fDvV+AV5+AQz4CH78OKkenXZUkSZJ6OEOb1JHGRlj4a5j3TSgrh7NvgPde4OiaJEmS9glDm9Set1+Dey+D1X+GiR+Cs34Og8ekXZUkSZJKiKFNak2M8MwtMPcaiI1Jo5Gpn3V0TZIkSfucoU1qbutamHM5rJoH406Es6+HoePSrkqSJEklytAmNYkRnrsD/vRVaKiDM34E7/snKCtLuzJJkiSVMEObBFCzHu7/Mqx4EA48Dmb8AoZPTLsqSZIkydCmEhcjLPkDPPgVqNsBH/4eHDcr6RIpSZIkFQBDm0rX9o3wwJWw7F6omgozboQRh6ZdlSRJkpTH0KbStGwO3H8F1G6FU78Jx18O5f5zkCRJUuHxKlWlZcdm+NNV8PzdsP9R8In7YNThaVclSZIktcnQptKx4iG473LYsQlOvhpOvBLKe6VdlSRJktQuQ5t6vp1vw9yrYfFtMHIKXHA3jH5P2lVJkiRJnWJoU8+2an6yUHbNOjjxK3DSVVDRJ+2qJEmSpE4ztKlnqq2Bh78Oi34L+02CS+clHSIlSZKkImNoU8/z8p/h3stgy+tJV8hTroFefdOuSpIkSdorhjb1HHXbYd434elfwrCJcPFcGDs97aokSZKkd8XQpp7h1Sdh9ix4azVM/3yy9lrv/mlXJUmSJL1rhjYVt907YcG/wZM3wJCx8JkHYNwH0q5KkiRJ6jKGNhW2ht1Q8yZseQO2Zramx1vWwFuvwK63YdrFcNp3oc/AtCuWJEmSupShTelpbEha8TcFsK1rcx6/kXxesw6I+c/rUwmVVTC4CkYfBUd8EiacnMIPIEmSJHU/Q5u6R2MjbK/OjIplAllTGNvSFMjehNiQ/7xeA5IwVlkFEw/LPh5cBZVjoPIA6FuZzs8kSZIkpcDQpj0XI2zfmISxpgD2zuOm7U1o3J3/vIq+2QA2/sT8MDa4KhPIhkAI6fxckiRJUgEytClfjLDzrZwpirn3kL2RnbbYUJv/vPLeSeiqHAMHHpczQjYm+VhZBf2HGcgkSZKkPWRoKyUxwq4tzQLYG9kpjE2jZvU7859XVgGDDkiCWNUxcNjHs2GsKZz13w/KytL5uSRJkqQezNDWk9TWNLuH7I1m0xbXQt22/OeEMhg0Ogle+x8Jk87ID2OVVTBwJJSVp/MzSZIkSSXO0FYs6na03WWxaYSsdkuzJwUYOCoJYCMmwcRT88PY4CoYuD+U+5+BJEmSVKi8Wi8Eu3flNPBo3mUxs+18q+XzBoxI7iMbNiGnsUemw2JlVTKCVtF73/88kiRJkrpMp0JbCOF04DqgHPhVjPEHzb7eB7gFmApsAj4dY3yla0stUvV1ULO29S6LTaNmOza2fF6/YdkQduD0/Jb3g6uSe8x69d33P48kSZKkfarD0BZCKAduAE4D1gBPhxDmxBiX5Rx2CfBWjPHgEMJ5wA+BT3dHwQWloT5Za6y9xaG3VdNicei+g7MBrOqY/Jb3Tft790/lR5IkSZJUWDoz0nYssCrG+DJACOEO4GwgN7SdDXwr8/ge4PoQQogxNksrRaSxAbatb6exx1rYtg5iY/7zeg/M3i826oiclvcHZB/3GZjOzyRJkiSp6HQmtFUBr+d8vgaY3tYxMcb6EMIWYDjQyry/AvbEz2D5/Ukgq3kTGuvzv17RL9vIY+IpLbssDq5KRtEkSZIkqYt0JrS1thpy8xG0zhxDCGEmMBNg7NixnXjpfWz3jmSR6INOaBnGKqug31AXh5YkSZK0T3UmtK0BDsz5fAywto1j1oQQKoDBwObm3yjGeDNwM8C0adMKb+rkyf+adgWSJEmSlKesE8c8DRwSQhgfQugNnAfMaXbMHOAfM48/BSwo6vvZJEmSJKlAdDjSlrlH7TJgLknL/9/EGJeGEL4DLIwxzgF+DdwaQlhFMsJ2XncWLUmSJEmlolPrtMUYHwQebLbvGzmPdwHndm1pkiRJkqTOTI+UJEmSJKXE0CZJkiRJBczQJkmSJEkFzNAmSZIkSQXM0CZJkiRJBczQJkmSJEkFzNAmSZIkSQXM0CZJkiRJBczQJkmSJEkFzNAmSZIkSQXM0CZJkiRJBczQJkmSJEkFzNAmSZIkSQXM0CZJkiRJBczQJkmSJEkFLMQY03nhEDYAr6by4u3bD9iYdhHaI56z4uM5Ky6er+LjOSs+nrPi4vkqPoV6zg6KMY7o6KDUQluhCiEsjDFOS7sOdZ7nrPh4zoqL56v4eM6Kj+esuHi+ik+xnzOnR0qSJElSATO0SZIkSVIBM7S1dHPaBWiPec6Kj+esuHi+io/nrPh4zoqL56v4FPU58542SZIkSSpgjrRJkiRJUgEztOUIIZweQlgRQlgVQvjXtOtR+0IIvwkhVIcQlqRdizoWQjgwhPBoCOGFEMLSEMKX0q5J7Qsh9A0hPBVCeC5zzr6ddk3qWAihPITwbAjh/rRrUcdCCK+EEJ4PISwOISxMux51LIQwJIRwTwhheeb/ae9Puya1LYQwKfPvq2nbGkL4ctp17SmnR2aEEMqBF4HTgDXA08D5McZlqRamNoUQPghsA26JMR6Rdj1qXwhhNDA6xvhMCGEQsAiY4b+xwhVCCMCAGOO2EEIv4P8CX4ox/k/KpakdIYQrgWlAZYzxzLTrUftCCK8A02KMhbh+lFoRQvgd8JcY469CCL2B/jHGt9OuSx3LXO+/AUyPMRbietFtcqQt61hgVYzx5RhjHXAHcHbKNakdMcbHgc1p16HOiTG+GWN8JvO4BngBqEq3KrUnJrZlPu2V2Xynr4CFEMYAHwN+lXYtUk8UQqgEPgj8GiDGWGdgKyqnAi8VW2ADQ1uuKuD1nM/X4AWl1C1CCOOAo4G/pluJOpKZarcYqAbmxRg9Z4Xtp8BVQGPahajTIvBwCGFRCGFm2sWoQxOADcD/yUxD/lUIYUDaRanTzgNuT7uIvWFoywqt7PMdZamLhRAGAn8Avhxj3Jp2PWpfjLEhxvheYAxwbAjBqcgFKoRwJlAdY1yUdi3aIyfEGI8BzgC+kJn6r8JVARwD/GeM8WhgO2AfhCKQmcp6FnB32rXsDUNb1hrgwJzPxwBrU6pF6pEy90X9AbgtxvjfadejzstM/3kMOD3lUtS2E4CzMvdI3QF8KITwX+mWpI7EGNdmPlYDfyS5XUOFaw2wJmfWwT0kIU6F7wzgmRjj+rQL2RuGtqyngUNCCOMzSfw8YE7KNUk9Rqapxa+BF2KM/5F2PepYCGFECGFI5nE/4O+A5elWpbbEGL8WYxwTYxxH8v+wBTHGC1MuS+0IIQzINGYiM8Xuw4AdkQtYjHEd8HoIYVJm16mADbWKw/kU6dRISIZ4BcQY60MIlwFzgXLgNzHGpSmXpXaEEG4HTgb2CyGsAb4ZY/x1ulWpHScAFwHPZ+6RArg6xvhgijWpfaOB32W6bZUBd8UYbSMvdZ1RwB+T97SoAH4fY3wo3ZLUCV8Ebsu8yf8y8NmU61EHQgj9STrEfy7tWvaWLf8lSZIkqYA5PVKSJEmSCpihTZIkSZIKmKFNkiRJkgqYoU2SJEmSCpihTZIkSZIKmKFNkiRJkgqYoU2SJEmSCpihTZIkSZIK2P8H9rpfIPu3zx0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5bdacb160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,8))\n",
    "plt.title(\"training and validation accuracies, depending on the epochs :\")\n",
    "x1 = df_evaluation['epochs_accuracies'][0]\n",
    "x2 = df_evaluation['epochs_val_accuracies'][0]\n",
    "y = range(0,len(x1))\n",
    "\n",
    "plt.plot(y,x1,label='training accuracy')\n",
    "plt.plot(y,x2,label='validation accuracy')\n",
    "plt.legend(loc='upper right')\n",
    "#The training accuracy is superior than the validation one : this is a good thing too.\n",
    "#But the training accuracy bursts the validation one : the gap is too important so there is a problem in the learning steps\n",
    "\n",
    "##Problem fixed : there was a problem with the cross_validation parameters in keras; by specifying the validation data the\n",
    "###learning steps were quite better (look at the notebook \"part_8_CNN_approach_results_analyzis_transfer_learning\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}